Automated Prediction of Proximal Middle Cerebral Artery Occlusions in Noncontrast Brain Computed Tomography.

BACKGROUND: Early identification of large vessel occlusion (LVO) in patients with ischemic stroke is crucial for timely interventions. We propose a machine learning-based algorithm (JLK-CTL) that uses handcrafted features from noncontrast computed tomography to predict LVO. METHODS: We included patients with ischemic stroke who underwent concurrent noncontrast computed tomography and computed tomography angiography in seven hospitals. Patients from 5 of these hospitals, admitted between May 2011 and March 2015, were randomly divided into training and internal validation (9:1 ratio). Those from the remaining 2 hospitals, admitted between March 2021 and September 2021, were designated for external validation. From each noncontrast computed tomography scan, we extracted differences in volume, tissue density, and Hounsfield unit distribution between bihemispheric regions (striatocapsular, insula, M1-M3, and M4-M6, modified from the Alberta Stroke Program Early Computed Tomography Score). A deep learning algorithm was used to incorporate clot signs as an additional feature. Machine learning models, including ExtraTrees, random forest, extreme gradient boosting, support vector machine, and multilayer perceptron, as well as a deep learning model, were trained and evaluated. Additionally, we assessed the models' performance after incorporating the National Institutes of Health Stroke Scale scores as an additional feature. RESULTS: Among 2919 patients, 83 were excluded. Across the training (n=2463), internal validation (n=275), and external validation (n=95) datasets, the mean ages were 68.5±12.4, 67.6±13.8, and 67.9±13.6 years, respectively. The proportions of men were 57%, 53%, and 59%, with LVO prevalences of 17.0%, 16.4%, and 26.3%, respectively. In the external validation, the ExtraTrees model achieved a robust area under the curve of 0.888 (95% CI, 0.850-0.925), with a sensitivity of 80.1% (95% CI, 72.0-88.1) and a specificity of 88.6% (95% CI, 84.7-92.5). Adding the National Institutes of Health Stroke Scale score to the ExtraTrees model increased sensitivity (from 80.1% to 92.1%) while maintaining specificity. CONCLUSIONS: Our algorithm provides reliable predictions of LVO using noncontrast computed tomography. By enabling early LVO identification, our algorithm has the potential to expedite the stroke workflow.

MRI Scoring Systems for Predicting Isocitrate Dehydrogenase Mutation and Chromosome 1p/19q Codeletion in Adult-type Diffuse Glioma Lacking Contrast Enhancement.

Background According to 2021 World Health Organization criteria, adult-type diffuse gliomas include glioblastoma, isocitrate dehydrogenase (IDH)-wildtype; oligodendroglioma, IDH-mutant and 1p/19q-codeleted; and astrocytoma, IDH-mutant, even when contrast enhancement is lacking. Purpose To develop and validate simple scoring systems for predicting IDH and subsequent 1p/19q codeletion status in gliomas without contrast enhancement using standard clinical MRI sequences. Materials and Methods This retrospective study included adult-type diffuse gliomas lacking contrast at contrast-enhanced MRI from two tertiary referral hospitals between January 2012 and April 2022 with diagnoses confirmed at pathology. IDH status was predicted primarily by using T2-fluid-attenuated inversion recovery (FLAIR) mismatch sign, followed by 1p/19q codeletion prediction. A visual rating of MRI features, apparent diffusion coefficient (ADC) ratio, and relative cerebral blood volume was measured. Scoring systems were developed through univariable and multivariable logistic regressions and underwent calibration and discrimination, including internal and external validation. Results For the internal validation cohort, 237 patients were included (mean age, 44.4 years ± 14.4 [SD]; 136 male patients; 193 patients in IDH prediction and 163 patients in 1p/19q prediction). For the external validation cohort, 35 patients were included (46.1 years ± 15.3; 20 male patients; 28 patients in IDH prediction and 24 patients in 1p/19q prediction). The T2-FLAIR mismatch sign demonstrated 100% specificity and 100% positive predictive value for IDH mutation. IDH status prediction scoring system for tumors without mismatch sign included age, ADC ratio, and morphologic characteristics, whereas 1p/19q codeletion prediction for IDH-mutant gliomas included ADC ratio, cortical involvement, and mismatch sign. For IDH status and 1p/19q codeletion prediction, bootstrap-corrected areas under the receiver operating characteristic curve were 0.86 (95% CI: 0.81, 0.90) and 0.73 (95% CI: 0.65, 0.81), respectively, whereas at external validation they were 0.99 (95% CI: 0.98, 1.0) and 0.88 (95% CI: 0.63, 1.0). Conclusion The T2-FLAIR mismatch sign and scoring systems using standard clinical MRI predicted IDH and 1p/19q codeletion status in gliomas lacking contrast enhancement. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Badve and Hodges in this issue.

Assessment of imaging risks for recurrence after stereotactic radiosurgery for brain metastases (IRRaS-BM).

BACKGROUND: The identification of viable tumors and radiation necrosis after stereotactic radiosurgery (SRS) is crucial for patient management. Tumor habitat analysis involving the grouping of similar voxels can identify subregions that share common biology and enable the depiction of areas of tumor recurrence and treatment-induced change. This study aims to validate an imaging biomarker for tumor recurrence after SRS for brain metastasis by conducting tumor habitat analysis using multi-parametric MRI. METHODS: In this prospective study (NCT05868928), patients with brain metastases will undergo multi-parametric MRI before SRS, and then follow-up MRIs will be conducted every 3 months until 24 months after SRS. The multi-parametric MRI protocol will include T2-weighted and contrast-enhanced T1-weighted imaging, diffusion-weighted imaging, and dynamic susceptibility contrast imaging. Using k-means voxel-wise clustering, this study will define three structural MRI habitats (enhancing, solid low-enhancing, and nonviable) on T1- and T2-weighted images and three physiologic MRI habitats (hypervascular cellular, hypovascular cellular, and nonviable) on apparent diffusion coefficient maps and cerebral blood volume maps. Using RANO-BM criteria as the reference standard, via Cox proportional hazards analysis, the study will prospectively evaluate associations between parameters of the tumor habitats and the time to recurrence. The DICE similarity coefficients between the recurrence site and tumor habitats will be calculated. DISCUSSION: The tumor habitat analysis will provide an objective and reliable measure for assessing tumor recurrence from brain metastasis following SRS. By identifying subregions for local recurrence, our study could guide the next therapeutic targets for patients after SRS. TRIAL REGISTRATION: This study is registered at ClinicalTrials.gov (NCT05868928).

Deep learning-based detection and quantification of brain metastases on black-blood imaging can provide treatment suggestions: a clinical cohort study.

OBJECTIVES: We aimed to evaluate whether deep learning-based detection and quantification of brain metastasis (BM) may suggest treatment options for patients with BMs. METHODS: The deep learning system (DLS) for detection and quantification of BM was developed in 193 patients and applied to 112 patients that were newly detected on black-blood contrast-enhanced T1-weighted imaging. Patients were assigned to one of 3 treatment suggestion groups according to the European Association of Neuro-Oncology (EANO)-European Society for Medical Oncology (ESMO) recommendations using number and volume of the BMs detected by the DLS: short-term imaging follow-up without treatment (group A), surgery or stereotactic radiosurgery (limited BM, group B), or whole-brain radiotherapy or systemic chemotherapy (extensive BM, group C). The concordance between the DLS-based groups and clinical decisions was analyzed with or without consideration of targeted agents. The performance of distinguishing high-risk (B + C) was calculated. RESULTS: Among 112 patients (mean age 64.3 years, 63 men), group C had the largest number and volume of BM, followed by group B (4.4 and 851.6 mm3) and A (1.5 and 15.5 mm3). The DLS-based groups were concordant with the actual clinical decisions, with an accuracy of 76.8% (86 of 112). Modified accuracy considering targeted agents was 81.3% (91 of 112). The DLS showed 95% (82/86) sensitivity and 81% (21/26) specificity for distinguishing the high risk. CONCLUSION: DLS-based detection and quantification of BM have the potential to be helpful in the determination of treatment options for both low- and high-risk groups of limited and extensive BMs. CLINICAL RELEVANCE STATEMENT: For patients with newly diagnosed brain metastasis, deep learning-based detection and quantification may be used in clinical settings where prompt and accurate treatment decisions are required, which can lead to better patient outcomes. KEY POINTS: • Deep learning-based brain metastasis detection and quantification showed excellent agreement with ground-truth classifications. • By setting an algorithm to suggest treatment based on the number and volume of brain metastases detected by the deep learning system, the concordance was 81.3%. • When dividing patients into low- and high-risk groups, the sensitivity for detecting the latter was 95%.

Achieving imaging and computational reproducibility on multiparametric MRI radiomics features in brain tumor diagnosis: phantom and clinical validation.

OBJECTIVES: The Image Biomarker Standardization Initiative has helped improve the computational reproducibility of MRI radiomics features. Nonetheless, the MRI sequences and features with high imaging reproducibility are yet to be established. To determine reproducible multiparametric MRI radiomics features across test-retest, multi-scanner, and computational reproducibility comparisons, and to evaluate their clinical value in brain tumor diagnosis. METHODS: To assess reproducibility, T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and diffusion-weighted imaging (DWI) were acquired from three 3-T MRI scanners using standardized phantom, and radiomics features were extracted using two computational algorithms. Reproducible radiomics features were selected when the concordance correlation coefficient value above 0.9 across multiple sessions, scanners, and computational algorithms. Random forest classifiers were trained with reproducible features (n = 117) and validated in a clinical cohort (n = 50) to evaluate whether features with high reproducibility improved the differentiation of glioblastoma from primary central nervous system lymphomas (PCNSLs). RESULTS: Radiomics features from T2WI demonstrated higher repeatability (65-94%) than those from DWI (38-48%) or T1WI (2-92%). Across test-retest, multi-scanner, and computational comparisons, T2WI provided 41 reproducible features, DWI provided six, and T1WI provided two. The performance of the classification model with reproducible features was higher than that using non-reproducible features in both training set (AUC, 0.916 vs. 0.877) and validation set (AUC, 0.957 vs. 0.869). CONCLUSION: Radiomics features with high reproducibility across multiple sessions, scanners, and computational algorithms were identified, and they showed higher diagnostic performance than non-reproducible radiomics features in the differentiation of glioblastoma from PCNSL. CLINICAL RELEVANCE STATEMENT: By identifying the radiomics features showing higher multi-machine reproducibility, our results also demonstrated higher radiomics diagnostic performance in the differentiation of glioblastoma from PCNSL, paving the way for further research designs and clinical application in neuro-oncology. KEY POINTS: • Highly reproducible radiomics features across multiple sessions, scanners, and computational algorithms were identified using phantom and applied to clinical diagnosis. • Radiomics features from T2-weighted imaging were more reproducible than those from T1-weighted and diffusion-weighted imaging. • Radiomics features with good reproducibility had better diagnostic performance for brain tumors than features with poor reproducibility.

Palsy of Both the Tibial Nerve and Common Peroneal Nerve Caused by a Ganglion Cyst in the Popliteal Area.

A ganglion cyst is a benign mass consisting of high-viscosity mucinous fluid. It can originate from the sheath of a tendon, peripheral nerve, or joint capsule. Compressive neuropathy caused by a ganglion cyst is rarely reported, with the majority of documented cases involving peroneal nerve palsy. To date, cases demonstrating both peroneal and tibial nerve palsies resulting from a ganglion cyst forming on a branch of the sciatic nerve have not been reported. In this paper, we present the case of a 74-year-old man visiting an outpatient clinic complaining of left-sided foot drop and sensory loss in the lower extremity, a lack of strength in his left leg, and a decrease in sensation in the leg for the past month without any history of trauma. Ankle dorsiflexion and great toe extension strength on the left side were Grade I. Ankle plantar flexion and great toe flexion were Grade II. We suspected peroneal and tibial nerve palsy and performed a screening ultrasound, which is inexpensive and rapid. In the operative field, several cysts were discovered, originating at the site where the sciatic nerve splits into peroneal and tibial nerves. After successful surgical decompression and a series of rehabilitation procedures, the patient's neurological symptoms improved. There was no recurrence.

Factors Influencing Wound Healing in Diabetic Foot Patients.

Background and objectives: Diabetic foot stands out as one of the most consequential and devastating complications of diabetes. Many factors, including VIPS (Vascular management, Infection management, Pressure relief, and Source of healing), influence the prognosis and treatment of diabetic foot patients. There are many studies on VIPS, but relatively few studies on "sources of healing". Nutrients that affect wound healing are known, but objective data in diabetic foot patients are insufficient. We hypothesized that "sources of healing" would have many effects on wound healing. The purpose of this study is to know the affecting factors related to the source of healing for diabetic foot patients. Materials and Methods: A retrospective review identified 46 consecutive patients who were admitted for diabetic foot management from July 2019 to April 2021 at our department. Several laboratory tests were performed for influencing factor evaluation. We checked serum levels of total protein, albumin, vitamin B, iron, zinc, magnesium, copper, Hb, HbA1c, HDL cholesterol, and LDL cholesterol. These values of diabetic foot patients were compared with normal values. Patients were divided into two groups based on wound healing rate, age, length of hospital stay, and sex, and the test values between the groups were compared. Results: Levels of albumin (37%) and Hb (89%) were low in the diabetic foot patients. As for trace elements, levels of iron (97%) and zinc (95%) were low in the patients, but levels of magnesium and copper were usually normal or high. There were no differences in demographic characteristics based on wound healing rate. However, when compared to normal adult values, diabetic foot patients in our data exhibited significantly lower levels of hemoglobin, total protein, albumin, iron, zinc, copper, and HDL cholesterol. When compared based on age and length of hospital stay, hemoglobin levels were significantly lower in both the older age group and the group with longer hospital stays. Conclusions: Serum levels of albumin, Hb, iron, and zinc were very low in most diabetic foot patients. These low values may have a negative relationship with wound healing. Nutrient replacements are necessary for wound healing in diabetic foot patients.

Reducing false positives in deep learning-based brain metastasis detection by using both gradient-echo and spin-echo contrast-enhanced MRI: validation in a multi-center diagnostic cohort.

OBJECTIVES: To develop a deep learning (DL) for detection of brain metastasis (BM) that incorporates both gradient- and turbo spin-echo contrast-enhanced MRI (dual-enhanced DL) and evaluate it in a clinical cohort in comparison with human readers and DL using gradient-echo-based imaging only (GRE DL). MATERIALS AND METHODS: DL detection was developed using data from 200 patients with BM (training set) and tested in 62 (internal) and 48 (external) consecutive patients who underwent stereotactic radiosurgery and diagnostic dual-enhanced imaging (dual-enhanced DL) and later guide GRE imaging (GRE DL). The detection sensitivity and positive predictive value (PPV) were compared between two DLs. Two neuroradiologists independently analyzed BM and reference standards for BM were separately drawn by another neuroradiologist. The relative differences (RDs) from the reference standard BM numbers were compared between the DLs and neuroradiologists. RESULTS: Sensitivity was similar between GRE DL (93%, 95% confidence interval [CI]: 90-96%) and dual-enhanced DL (92% [89-94%]). The PPV of the dual-enhanced DL was higher (89% [86-92%], p < .001) than that of GRE DL (76%, [72-80%]). GRE DL significantly overestimated the number of metastases (false positives; RD: 0.05, 95% CI: 0.00-0.58) compared with neuroradiologists (RD: 0.00, 95% CI: - 0.28, 0.15, p < .001), whereas dual-enhanced DL (RD: 0.00, 95% CI: 0.00-0.15) did not show a statistically significant difference from neuroradiologists (RD: 0.00, 95% CI: - 0.20-0.10, p = .913). CONCLUSION: The dual-enhanced DL showed improved detection of BM and reduced overestimation compared with GRE DL, achieving similar performance to neuroradiologists. CLINICAL RELEVANCE STATEMENT: The use of deep learning-based brain metastasis detection with turbo spin-echo imaging reduces false positive detections, aiding in the guidance of stereotactic radiosurgery when gradient-echo imaging alone is employed. KEY POINTS: •Deep learning for brain metastasis detection improved by using both gradient- and turbo spin-echo contrast-enhanced MRI (dual-enhanced deep learning). •Dual-enhanced deep learning increased true positive detections and reduced overestimation. •Dual-enhanced deep learning achieved similar performance to neuroradiologists for brain metastasis counts.

Vessel size and perfusion-derived vascular habitat refines prediction of treatment failure to bevacizumab in recurrent glioblastomas: validation in a prospective cohort.

OBJECTIVES: Anti-angiogenic therapy may not benefit all patients with recurrent glioblastomas, and imaging biomarker predicting treatment response to anti-angiogenic therapy is currently limited. We aimed to develop and validate vascular habitats based on perfusion and vessel size to predict time to progression (TTP) in patients with recurrent glioblastomas treated with bevacizumab. METHODS: Sixty-nine patients with recurrent glioblastomas treated with bevacizumab who underwent pretreatment MRI with dynamic susceptibility contrast imaging and vessel architectural imaging were enrolled. Vascular habitats were constructed using vessel size index (VSI) and relative cerebral blood volume (rCBV). Associations with vascular habitats and TTP were analyzed using Cox proportional hazard regression analysis. In a prospectively enrolled validation cohort consisting of 15 patients ( ClinicalTrials.gov identifier; NCT04143425), stratification of TTP was demonstrated by the Kaplan-Meier method (log-rank test) using vascular habitats. RESULTS: Three vascular habitats consisting of high, intermediate, and low angiogenic habitats were identified with rCBV and VSI. Both high angiogenic and intermediate angiogenic habitats were significantly associated with a shorter TTP (hazard ratio [HR], 2.78 and 1.82, respectively; largest p = .003) and so was rCBV (HR, 2.15; p = .02). Concordance probability index of vascular habitat combining high and intermediate angiogenic habitats was 0.74. Vascular habitats stratified patients as good or poor responder in a prospective cohort (p = .059). CONCLUSIONS: Perfusion- and vessel size-derived vascular habitats predicted TTP in recurrent glioblastomas treated with anti-angiogenic therapy and aided patient stratification in a prospective validation cohort. TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT04143425 KEY POINTS: • High and intermediate angiogenic habitats predicted TTP in recurrent glioblastomas treated with anti-angiogenic therapy. • Vascular habitats combining high and intermediate angiogenic habitats aided patient stratification for anti-angiogenic therapy in recurrent glioblastomas.

Defining subventricular zone involvement to predict the survival of patients in isocitrate dehydrogenase-wild type glioblastoma: validation in a prospective registry.

OBJECTIVES: The prognostic value of subventricular zone distance (SVD) is unclear because of different definitions and lack of evaluation of clinical survival models. The aim of this study was to define SVD and evaluate its prognostic value in a survival nomogram for glioblastoma. METHODS: This retrospective study included 158 (SVD biomarker) from historical glioblastoma patients and 187 (survival modeling) with IDH-wild type glioblastoma from a prospective registry (NCT02619890). SVD was assessed by two radiologists: definition 1, the distance between the tumor edge to subventricular zone (SVZ); definition 2, the distance between the tumor centroid to SVZ; definition 3, enhancement at the ventricular wall. The associations between SVD and overall survival (OS) were evaluated using multivariable Cox proportional hazards regression analysis. Performance of an updated SVD survival model was compared with that of the Radiation Therapy Oncology Group (RTOG) 0525 nomogram. RESULTS: SVD according to both definition 1 (hazard ratio [HR]: 0.97, 95% CI: 0.94-0.99; p = .011) and definition 2 (HR: 0.96, 0.94-0.98, p < .001) was adversely associated with OS. Definition 1 was adversely associated with PFS (HR: 0.96, 0.94-0.99, p = .008) and showed the highest reproducibility (intraclass correlation coefficient, 0.90). The SVD-updated model showed similar to better performance than the RTOG model for predicting OS of up to 3 years (AUC: 0.735-0.738 vs. 0.687-0.708), with higher time-dependent specificity for 1-year (89.9% vs. 70.6%) and 3-year OS (93.3% vs. 80.0%). CONCLUSION: SVZ distance is an independent adverse prognostic factor in patients with IDH-wild type glioblastoma. Updating the survival model with SVZ provides better time-dependent specificity and reproducibility. KEY POINTS: • Subventricular zone distance (SVD) measurement from tumor edge showed high reproducibility. • Longer SVD was independently associated with longer overall survival. • Adding SVD improved time-dependent specificity for survival model in a prospective registry.

Prediction of pseudoprogression in post-treatment glioblastoma using dynamic susceptibility contrast-derived oxygenation and microvascular transit time heterogeneity measures.

OBJECTIVES: To evaluate the added value of MR dynamic susceptibility contrast (DSC)-perfusion-weighted imaging (PWI)-derived tumour microvascular and oxygenation information with cerebral blood volume (CBV) to distinguish pseudoprogression from true progression (TP) in post-treatment glioblastoma. METHODS: This retrospective single-institution study included patients with isocitrate dehydrogenase (IDH) wild-type glioblastoma and a newly developed or enlarging measurable contrast-enhancing mass within 12 weeks after concurrent chemoradiotherapy. CBV, capillary transit time heterogeneity (CTH), oxygen extraction fraction (OEF), and cerebral metabolic rate of oxygen (CMRO2) were obtained from DSC-PWI. Predictors were selected using univariable logistic regression, and performance was measured with adjusted diagnostic odds with tumour volume and area under the curve (AUC) of receiver operating characteristics analysis. RESULTS: A total of 103 patients were included (mean age, 59.6 years; 59 women), with 67 cases of TP and 36 cases of pseudoprogression. Pseudoprogression exhibited higher CTH (4.0 vs. 3.4, p = .019) and higher OEF (12.7 vs. 10.7, p = .014) than TP, but a similar CBV (1.48 vs. 1.53, p = .13) and CMRO2 (7.7 vs. 7.3s, p = .598). Independent of tumour volume, both high CTH (adjusted odds ratio [OR] 1.52; 95% confidence interval [CI]: 1.11-2.09, p = .009) and high OEF (adjusted OR 1.17; 95% CI:1.03-1.33, p = .016) were predictors of pseudoprogression. The combination of CTH, OEF, and CBV yielded higher diagnostic performance (AUC 0.71) than CBV alone (AUC 0.65). CONCLUSION: High intratumoural capillary transit heterogeneity and high oxygen extraction fraction derived from DSC-PWI have enhanced the diagnostic value of CBV in pseudoprogression of post-treatment IDH-wild type glioblastoma. CLINICAL RELEVANCE STATEMENT: In the early post-treatment stage of glioblastoma, pseudoprogression exhibited both high oxygen extraction fraction and high capillary transit heterogeneity and these dynamic susceptibility contrast-perfusion weighted imaging derived parameters have added value in cerebral blood volume-based noninvasive differentiation of pseudoprogression from true progression. KEY POINTS: • Capillary transit time heterogeneity and oxygen extraction fraction can be measured noninvasively through processing of dynamic susceptibility contrast imaging. • Pseudoprogression exhibited higher capillary transit time heterogeneity and higher oxygen extraction fraction than true progression. • A combination of cerebral blood volume, capillary transit time heterogeneity, and oxygen extraction fraction yielded the highest diagnostic performance (area under the curve 0.71).

Development and validation of an automatic classification algorithm for the diagnosis of Alzheimer's disease using a high-performance interpretable deep learning network.

OBJECTIVES: To develop and validate an automatic classification algorithm for diagnosing Alzheimer's disease (AD) or mild cognitive impairment (MCI). METHODS AND MATERIALS: This study evaluated a high-performance interpretable network algorithm (TabNet) and compared its performance with that of XGBoost, a widely used classifier. Brain segmentation was performed using a commercially approved software. TabNet and XGBoost were trained on the volumes or radiomics features of 102 segmented regions for classifying subjects into AD, MCI, or cognitively normal (CN) groups. The diagnostic performances of the two algorithms were compared using areas under the curves (AUCs). Additionally, 20 deep learning-based AD signature areas were investigated. RESULTS: Between December 2014 and March 2017, 161 AD, 153 MCI, and 306 CN cases were enrolled. Another 120 AD, 90 MCI, and 141 CN cases were included for the internal validation. Public datasets were used for external validation. TabNet with volume features had an AUC of 0.951 (95% confidence interval [CI], 0.947-0.955) for AD vs CN, which was similar to that of XGBoost (0.953 [95% CI, 0.951-0.955], p = 0.41). External validation revealed the similar performances of two classifiers using volume features (0.871 vs. 0.871, p = 0.86). Likewise, two algorithms showed similar performances with one another in classifying MCI. The addition of radiomics data did not improve the performance of TabNet. TabNet and XGBoost focused on the same 13/20 regions of interest, including the hippocampus, inferior lateral ventricle, and entorhinal cortex. CONCLUSIONS: TabNet shows high performance in AD classification and detailed interpretation of the selected regions. CLINICAL RELEVANCE STATEMENT: Using a high-performance interpretable deep learning network, the automatic classification algorithm assisted in accurate Alzheimer's disease detection using 3D T1-weighted brain MRI and detailed interpretation of the selected regions. KEY POINTS: • MR volumetry data revealed that TabNet had a high diagnostic performance in differentiating Alzheimer's disease (AD) from cognitive normal cases, which was comparable with that of XGBoost. • The addition of radiomics data to the volume data did not improve the diagnostic performance of TabNet. • Both TabNet and XGBoost selected the clinically meaningful regions of interest in AD, including the hippocampus, inferior lateral ventricle, and entorhinal cortex.

Natural language processing to predict isocitrate dehydrogenase genotype in diffuse glioma using MR radiology reports.

OBJECTIVES: To evaluate the performance of natural language processing (NLP) models to predict isocitrate dehydrogenase (IDH) mutation status in diffuse glioma using routine MR radiology reports. MATERIALS AND METHODS: This retrospective, multi-center study included consecutive patients with diffuse glioma with known IDH mutation status from May 2009 to November 2021 whose initial MR radiology report was available prior to pathologic diagnosis. Five NLP models (long short-term memory [LSTM], bidirectional LSTM, bidirectional encoder representations from transformers [BERT], BERT graph convolutional network [GCN], BioBERT) were trained, and area under the receiver operating characteristic curve (AUC) was assessed to validate prediction of IDH mutation status in the internal and external validation sets. The performance of the best performing NLP model was compared with that of the human readers. RESULTS: A total of 1427 patients (mean age ± standard deviation, 54 ± 15; 779 men, 54.6%) with 720 patients in the training set, 180 patients in the internal validation set, and 527 patients in the external validation set were included. In the external validation set, BERT GCN showed the highest performance (AUC 0.85, 95% CI 0.81-0.89) in predicting IDH mutation status, which was higher than LSTM (AUC 0.77, 95% CI 0.72-0.81; p = .003) and BioBERT (AUC 0.81, 95% CI 0.76-0.85; p = .03). This was higher than that of a neuroradiologist (AUC 0.80, 95% CI 0.76-0.84; p = .005) and a neurosurgeon (AUC 0.79, 95% CI 0.76-0.84; p = .04). CONCLUSION: BERT GCN was externally validated to predict IDH mutation status in patients with diffuse glioma using routine MR radiology reports with superior or at least comparable performance to human reader. CLINICAL RELEVANCE STATEMENT: Natural language processing may be used to extract relevant information from routine radiology reports to predict cancer genotype and provide prognostic information that may aid in guiding treatment strategy and enabling personalized medicine. KEY POINTS: • A transformer-based natural language processing (NLP) model predicted isocitrate dehydrogenase mutation status in diffuse glioma with an AUC of 0.85 in the external validation set. • The best NLP models were superior or at least comparable to human readers in both internal and external validation sets. • Transformer-based models showed higher performance than conventional NLP model such as long short-term memory.

Deep learning referral suggestion and tumour discrimination using explainable artificial intelligence applied to multiparametric MRI.

OBJECTIVES: An appropriate and fast clinical referral suggestion is important for intra-axial mass-like lesions (IMLLs) in the emergency setting. We aimed to apply an interpretable deep learning (DL) system to multiparametric MRI to obtain clinical referral suggestion for IMLLs, and to validate it in the setting of nontraumatic emergency neuroradiology. METHODS: A DL system was developed in 747 patients with IMLLs ranging 30 diseases who underwent pre- and post-contrast T1-weighted (T1CE), FLAIR, and diffusion-weighted imaging (DWI). A DL system that segments IMLLs, classifies tumourous conditions, and suggests clinical referral among surgery, systematic work-up, medical treatment, and conservative treatment, was developed. The system was validated in an independent cohort of 130 emergency patients, and performance in referral suggestion and tumour discrimination was compared with that of radiologists using receiver operating characteristics curve, precision-recall curve analysis, and confusion matrices. Multiparametric interpretable visualisation of high-relevance regions from layer-wise relevance propagation overlaid on contrast-enhanced T1WI and DWI was analysed. RESULTS: The DL system provided correct referral suggestions in 94 of 130 patients (72.3%) and performed comparably to radiologists (accuracy 72.6%, McNemar test; p = .942). For distinguishing tumours from non-tumourous conditions, the DL system (AUC, 0.90 and AUPRC, 0.94) performed similarly to human readers (AUC, 0.81~0.92, and AUPRC, 0.88~0.95). Solid portions of tumours showed a high overlap of relevance, but non-tumours did not (Dice coefficient 0.77 vs. 0.33, p < .001), demonstrating the DL's decision. CONCLUSIONS: Our DL system could appropriately triage patients using multiparametric MRI and provide interpretability through multiparametric heatmaps, and may thereby aid neuroradiologic diagnoses in emergency settings. CLINICAL RELEVANCE STATEMENT: Our AI triages patients with raw MRI images to clinical referral pathways in brain intra-axial mass-like lesions. We demonstrate that the decision is based on the relative relevance between contrast-enhanced T1-weighted and diffusion-weighted images, providing explainability across multiparametric MRI data. KEY POINTS: • A deep learning (DL) system using multiparametric MRI suggested clinical referral to patients with intra-axial mass-like lesions (IMLLs) similar to radiologists (accuracy 72.3% vs. 72.6%). • In the differentiation of tumourous and non-tumourous conditions, the DL system (AUC, 0.90) performed similar with radiologists (AUC, 0.81-0.92). • The DL's decision basis for differentiating tumours from non-tumours can be quantified using multiparametric heatmaps obtained via the layer-wise relevance propagation method.

Risk estimation for idiopathic normal-pressure hydrocephalus: development and validation of a brain morphometry-based nomogram.

OBJECTIVES: To develop and validate a nomogram based on MRI features for predicting iNPH. METHODS: Patients aged ≥ 60 years (clinically diagnosed with iNPH, Parkinson's disease, or Alzheimer's disease or healthy controls) who underwent MRI including three-dimensional T1-weighted volumetric MRI were retrospectively identified from two tertiary referral hospitals (one hospital for derivation set and the other for validation set). Clinical and imaging features for iNPH were assessed. Deep learning-based brain segmentation software was used for 3D volumetry. A prediction model was developed using logistic regression and transformed into a nomogram. The performance of the nomogram was assessed with respect to discrimination and calibration abilities. The nomogram was internally and externally validated. RESULTS: A total of 452 patients (mean age ± SD, 73.2 ± 6.5 years; 200 men) were evaluated as the derivation set. One hundred eleven and 341 patients were categorized into the iNPH and non-iNPH groups, respectively. In multivariable analysis, high-convexity tightness (odds ratio [OR], 35.1; 95% CI: 4.5, 275.5), callosal angle < 90° (OR, 12.5; 95% CI: 3.1, 50.0), and normalized lateral ventricle volume (OR, 4.2; 95% CI: 2.7, 6.7) were associated with iNPH. The nomogram combining these three variables showed an area under the curve of 0.995 (95% CI: 0.991, 0.999) in the study sample, 0.994 (95% CI: 0.990, 0.998) in the internal validation sample, and 0.969 (95% CI: 0.940, 0.997) in the external validation sample. CONCLUSION: A brain morphometry-based nomogram including high-convexity tightness, callosal angle < 90°, and normalized lateral ventricle volume can help accurately estimate the probability of iNPH. KEY POINTS: • The nomogram with MRI findings (high-convexity tightness, callosal angle, and normalized lateral ventricle volume) helped in predicting the probability of idiopathic normal-pressure hydrocephalus. • The nomogram may facilitate the prediction of idiopathic normal-pressure hydrocephalus and consequently avoid unnecessary invasive procedures such as the cerebrospinal fluid tap test, drainage test, and cerebrospinal fluid shunt surgery.

Fully automated radiomics-based machine learning models for multiclass classification of single brain tumors: Glioblastoma, lymphoma, and metastasis.

BACKGROUND AND PURPOSE: To investigate the diagnostic performance of fully automated radiomics-based models for multiclass classification of a single enhancing brain tumor among glioblastoma, central nervous system lymphoma, and metastasis. MATERIALS AND METHODS: The training and test sets were comprised of 538 cases (300 glioblastomas, 73 lymphomas, and 165 metastases) and 169 cases (101 glioblastomas, 29 lymphomas, and 39 metastases), respectively. After fully automated segmentation, radiomic features were extracted. Three conventional machine learning classifiers, including least absolute shrinkage and selection operator (LASSO), adaptive boosting (Adaboost), and support vector machine with the linear kernel (SVC), combined with one of four feature selection methods, including forward sequential feature selection, F score, mutual information, and LASSO, were trained. Additionally, one ensemble classifier based on the three classifiers was used. The diagnostic performance of the optimized models was tested in the test set using the accuracy, F1-macro score, and the area under the receiver operating characteristic curve (AUCROC). RESULTS: The best performance was achieved when the LASSO was used as a feature selection method. In the test set, the best performance was achieved by the ensemble classifier, showing an accuracy of 76.3% (95% CI, 70.0-82.7), a F1-macro score of 0.704, and an AUCROC of 0.878. CONCLUSION: Our fully automated radiomics-based models for multiclass classification might be useful for differential diagnosis of a single enhancing brain tumor with a good diagnostic performance and generalizability.

Treatment of Soft Tissue Defects after Minimally Invasive Plate Osteosynthesis in Fractures of the Distal Tibia: Clinical Results after Reverse Sural Artery Flap.

Introduction: Distal tibial fractures make up approximately 3% to 10% of all tibial fractures or about 1% of lower extremity fractures. MIPO is an appropriate procedure and method to achieve stable metal plate fixation and osseointegration by minimizing soft tissue damage and vascular integrity at the fracture site. MIPO to the medial tibia during distal tibial fractures induces skin irritation due to the thickness of the metal plate, which causes discomfort and pain on the medial side of the distal leg, and if severe, complications such as infection and skin defect may occur. The reverse sural flap is a well-researched approach for covering defects in the lower third of the leg, ankle, and foot. Materials and Methods: Among 151 patients with distal tibia fractures who underwent minimally invasive metal plate fixation, soft tissue was injured due to postoperative complications. We treated 13 cases with necrosis and exposed metal plates by retrograde nasogastric artery flap surgery. For these patients, we collected obligatory patient records, radiological data, and wound photographs of the treatment results and complications of reconstructive surgery. Results: In all the cases, flap survival was confirmed at the final outpatient follow-up. The exposed area of the metal plate was well coated, and there was no plate failure due to complete necrosis. Three out of four women complained of aesthetic dissatisfaction because the volume of the tunnel through which the skin mirror passed and the skin plate itself were thick. In two cases, defatting was performed to reduce the thickness of the plate while removing the metal plate. Conclusions: Metal plate exposure after distal tibial fractures have been treated with minimally invasive metal plate fusion and can be successfully treated with retrograde nasogastric artery flaps, and several surgical techniques are used during flap surgery.

Contrast enhancing pattern on pre-treatment MRI predicts response to anti-angiogenic treatment in recurrent glioblastoma: comparison of bevacizumab and temozolomide treatment.

OBJECTIVE: To evaluate the value of the contrast enhancing pattern on pre-treatment MRI for predicting the response to anti-angiogenic treatment in patients with IDH-wild type recurrent glioblastoma. METHODS: This retrospective study enrolled 65 patients with IDH wild-type recurrent glioblastoma who received standard therapy and then received either bevacizumab (46 patients) or temozolomide (19 patients) as a secondary treatment. The contrast enhancing pattern on pre-treatment MRI was visually analyzed and dichotomized into contrast enhancing lesion (CEL) dominant and non-enhancing lesion (NEL) dominant types. Quantitative volumetric analysis was used to support the dichotomization. The Kaplan-Meier method and Cox proportional hazards regression analysis were used to stratify progression free survival (PFS) according to the treatment in the entire patients, CEL dominant group, and NEL dominant group. RESULTS: In all patients, the PFS of those treated with bevacizumab was not significantly different from those treated with temozolomide (log-rank test, P = 0.96). When the contrast enhancing pattern was considered, bevacizumab was associated with longer PFS in the CEL dominant group (P = 0.031), whereas temozolomide showed longer PFS in the NEL dominant group (P = 0.022). Quantitative analysis revealed mean values for the proportion of solid-enhancing tumor of 13.7% for the CEL dominant group and 4.3% for the NEL dominant group. CONCLUSION: Patients with the CEL dominant type showed a better treatment response to bevacizumab, whereas NEL dominant types showed a better response to temozolomide. The contrast enhancing pattern on pre-treatment MRI can be used to stratify patients with IDH wild-type recurrent glioblastoma according to the effect of anti-angiogenic treatment.

Tumor habitat analysis by magnetic resonance imaging distinguishes tumor progression from radiation necrosis in brain metastases after stereotactic radiosurgery.

OBJECTIVES: The identification of viable tumor after stereotactic radiosurgery (SRS) is important for future targeted therapy. This study aimed to determine whether tumor habitat on structural and physiologic MRI can distinguish viable tumor from radiation necrosis of brain metastases after SRS. METHOD: Multiparametric contrast-enhanced T1- and T2-weighted imaging, apparent diffusion coefficient (ADC), and cerebral blood volume (CBV) were obtained from 52 patients with 69 metastases, showing enlarging enhancing masses after SRS. Voxel-wise clustering identified three structural MRI habitats (enhancing, solid low-enhancing, and nonviable) and three physiologic MRI habitats (hypervascular cellular, hypovascular cellular, and nonviable). Habitat-based predictors for viable tumor or radiation necrosis were identified by logistic regression. Performance was validated using the area under the curve (AUC) of the receiver operating characteristics curve in an independent dataset with 24 patients. RESULTS: None of the physiologic MRI habitats was indicative of viable tumor. Viable tumor was predicted by a high-volume fraction of solid low-enhancing habitat (low T2-weighted and low CE-T1-weighted values; odds ratio [OR] 1.74, p <.001) and a low-volume fraction of nonviable tissue habitat (high T2-weighted and low CE-T1-weighted values; OR 0.55, p <.001). Combined structural MRI habitats yielded good discriminatory ability in both development (AUC 0.85, 95% confidence interval [CI]: 0.77-0.94) and validation sets (AUC 0.86, 95% CI:0.70-0.99), outperforming single ADC (AUC 0.64) and CBV (AUC 0.58) values. The site of progression matched with the solid low-enhancing habitat (72%, 8/11). CONCLUSION: Solid low-enhancing and nonviable tissue habitats on structural MRI can help to localize viable tumor in patients with brain metastases after SRS. KEY POINTS: • Structural MRI habitats helped to differentiate viable tumor from radiation necrosis. • Solid low-enhancing habitat was most helpful to find viable tumor. • Providing spatial information, the site of progression matched with solid low-enhancing habitat.

Diffuse glioma, not otherwise specified: imaging-based risk stratification achieves histomolecular-level prognostication.

OBJECTIVES: To determine whether imaging-based risk stratification enables prognostication in diffuse glioma, NOS (not otherwise specified). METHODS: Data from 220 patients classified as diffuse glioma, NOS, between January 2011 and December 2020 were retrospectively included. Two neuroradiologists analyzed pre-surgical CT and MRI to assign gliomas to the three imaging-based risk types considering well-known imaging phenotypes (e.g., T2/FLAIR mismatch). According to the 2021 World Health Organization classification, the three risk types included (1) low-risk, expecting oligodendroglioma, isocitrate dehydrogenase (IDH)-mutant, and 1p/19q-codeleted; (2) intermediate-risk, expecting astrocytoma, IDH-mutant; and (3) high-risk, expecting glioblastoma, IDH-wildtype. Progression-free survival (PFS) and overall survival (OS) were estimated for each risk type. Time-dependent receiver operating characteristic analysis using 10-fold cross-validation with 100-fold bootstrapping was used to compare the performance of an imaging-based survival model with that of a historical molecular-based survival model published in 2015, created using The Cancer Genome Archive data. RESULTS: Prognostication according to the three imaging-based risk types was achieved for both PFS and OS (log-rank test, p < 0.001). The imaging-based survival model showed high prognostic value, with areas under the curves (AUCs) of 0.772 and 0.650 for 1-year PFS and OS, respectively, similar to the historical molecular-based survival model (AUC = 0.74 for PFS and 0.87 for OS). The imaging-based survival model achieved high long-term performance in both 3-year PFS (AUC = 0.806) and 5-year OS (AUC = 0.812). CONCLUSION: Imaging-based risk stratification achieved histomolecular-level prognostication in diffuse glioma, NOS, and could aid in guiding patient referral for insufficient or unsuccessful molecular diagnosis. KEY POINTS: • Three imaging-based risk types enable distinct prognostication in diffuse glioma, NOS (not otherwise specified). • The imaging-based survival model achieved similar prognostic performance as a historical molecular-based survival model. • For long-term prognostication of 3 and 5 years, the imaging-based survival model showed high performance.