Peak detection in intracranial pressure signal waveforms: a comparative study.

BACKGROUND: The monitoring and analysis of quasi-periodic biological signals such as electrocardiography (ECG), intracranial pressure (ICP), and cerebral blood flow velocity (CBFV) waveforms plays an important role in the early detection of adverse patient events and contributes to improved care management in the intensive care unit (ICU). This work quantitatively evaluates existing computational frameworks for automatically extracting peaks within ICP waveforms. METHODS: Peak detection techniques based on state-of-the-art machine learning models were evaluated in terms of robustness to varying noise levels. The evaluation was performed on a dataset of ICP signals assembled from 700 h of monitoring from 64 neurosurgical patients. The groundtruth of the peak locations was established manually on a subset of 13, 611 pulses. Additional evaluation was performed using a simulated dataset of ICP with controlled temporal dynamics and noise. RESULTS: The quantitative analysis of peak detection algorithms applied to individual waveforms indicates that most techniques provide acceptable accuracy with a mean absolute error (MAE) ≤ 10 ms without noise. In the presence of a higher noise level, however, only kernel spectral regression and random forest remain below that error threshold while the performance of other techniques deteriorates. Our experiments also demonstrated that tracking methods such as Bayesian inference and long short-term memory (LSTM) can be applied continuously and provide additional robustness in situations where single pulse analysis methods fail, such as missing data. CONCLUSION: While machine learning-based peak detection methods require manually labeled data for training, these models outperform conventional signal processing ones based on handcrafted rules and should be considered for peak detection in modern frameworks. In particular, peak tracking methods that incorporate temporal information between successive periods of the signals have demonstrated in our experiments to provide more robustness to noise and temporary artifacts that commonly arise as part of the monitoring setup in the clinical setting.

Nodule detection and generation on chest X-rays: NODE21 Challenge.

Pulmonary nodules may be an early manifestation of lung cancer, the leading cause of cancer-related deaths among both men and women. Numerous studies have established that deep learning methods can yield high-performance levels in the detection of lung nodules in chest X-rays. However, the lack of gold-standard public datasets slows down the progression of the research and prevents benchmarking of methods for this task. To address this, we organized a public research challenge, NODE21, aimed at the detection and generation of lung nodules in chest X-rays. While the detection track assesses state-of-the-art nodule detection systems, the generation track determines the utility of nodule generation algorithms to augment training data and hence improve the performance of the detection systems. This paper summarizes the results of the NODE21 challenge and performs extensive additional experiments to examine the impact of the synthetically generated nodule training images on the detection algorithm performance.

AtPCa-Net: anatomical-aware prostate cancer detection network on multi-parametric MRI.

Multi-parametric MRI (mpMRI) is widely used for prostate cancer (PCa) diagnosis. Deep learning models show good performance in detecting PCa on mpMRI, but domain-specific PCa-related anatomical information is sometimes overlooked and not fully explored even by state-of-the-art deep learning models, causing potential suboptimal performances in PCa detection. Symmetric-related anatomical information is commonly used when distinguishing PCa lesions from other visually similar but benign prostate tissue. In addition, different combinations of mpMRI findings are used for evaluating the aggressiveness of PCa for abnormal findings allocated in different prostate zones. In this study, we investigate these domain-specific anatomical properties in PCa diagnosis and how we can adopt them into the deep learning framework to improve the model's detection performance. We propose an anatomical-aware PCa detection Network (AtPCa-Net) for PCa detection on mpMRI. Experiments show that the AtPCa-Net can better utilize the anatomical-related information, and the proposed anatomical-aware designs help improve the overall model performance on both PCa detection and patient-level classification.

Prediction of Visual Field Progression with Baseline and Longitudinal Structural Measurements Using Deep Learning.

PURPOSE: Identifying glaucoma patients at high risk of progression based on widely available structural data is an unmet task in clinical practice. We test the hypothesis that baseline or serial structural measures can predict visual field (VF) progression with deep learning (DL). DESIGN: Development of a DL algorithm to predict VF progression. METHODS: 3,079 eyes (1,765 patients) with various types of glaucoma and ≥5 VFs, and ≥3 years of follow-up from a tertiary academic center were included. Serial VF mean deviation (MD) rates of change were estimated with linear-regression. VF progression was defined as negative MD slope with p<0.05. A Siamese Neural Network with ResNet-152 backbone pre-trained on ImageNet was designed to predict VF progression using serial optic-disc photographs (ODP), and baseline retinal nerve fiber layer (RNFL) thickness. We tested the model on a separate dataset (427 eyes) with RNFL data from different OCT. The Main Outcome Measure was Area under ROC curve (AUC). RESULTS: Baseline average (SD) MD was 3.4 (4.9)dB. VF progression was detected in 900 eyes (29%). AUC (95% CI) for model incorporating baseline ODP and RNFL thickness was 0.813 (0.757-0.869). After adding the second and third ODPs, AUC increased to 0.860 and 0.894, respectively (p<0.027). This model also had highest AUC (0.911) for predicting fast progression (MD rate <1.0 dB/year). Model's performance was similar when applied to second dataset using RNFL data from another OCT device (AUC=0.893; 0.837-0.948). CONCLUSIONS: DL model predicted VF progression with clinically relevant accuracy using baseline RNFL thickness and serial ODPs and can be implemented as a clinical tool after further validation.

Clinical characteristics and outcomes of patients with post-stroke epilepsy: protocol for an individual patient data meta-analysis from the International Post-stroke Epilepsy Research Repository (IPSERR).

INTRODUCTION: Despite significant advances in managing acute stroke and reducing stroke mortality, preventing complications like post-stroke epilepsy (PSE) has seen limited progress. PSE research has been scattered worldwide with varying methodologies and data reporting. To address this, we established the International Post-stroke Epilepsy Research Consortium (IPSERC) to integrate global PSE research efforts. This protocol outlines an individual patient data meta-analysis (IPD-MA) to determine outcomes in patients with post-stroke seizures (PSS) and develop/validate PSE prediction models, comparing them with existing models. This protocol informs about creating the International Post-stroke Epilepsy Research Repository (IPSERR) to support future collaborative research. METHODS AND ANALYSIS: We utilised a comprehensive search strategy and searched MEDLINE, Embase, PsycInfo, Cochrane, and Web of Science databases until 30 January 2023. We extracted observational studies of stroke patients aged ≥18 years, presenting early or late PSS with data on patient outcome measures, and conducted the risk of bias assessment. We did not apply any restriction based on the date or language of publication. We will invite these study authors and the IPSERC collaborators to contribute IPD to IPSERR. We will review the IPD lodged within IPSERR to identify patients who developed epileptic seizures and those who did not. We will merge the IPD files of individual data and standardise the variables where possible for consistency. We will conduct an IPD-MA to estimate the prognostic value of clinical characteristics in predicting PSE. ETHICS AND DISSEMINATION: Ethics approval is not required for this study. The results will be published in peer-reviewed journals. This study will contribute to IPSERR, which will be available to researchers for future PSE research projects. It will also serve as a platform to anchor future clinical trials. TRIAL REGISTRATION NUMBER: NCT06108102.

Prediction of Central Visual Field Measures From Macular OCT Volume Scans With Deep Learning.

Purpose: Predict central 10° global and local visual field (VF) measurements from macular optical coherence tomography (OCT) volume scans with deep learning (DL). Methods: This study included 1121 OCT volume scans and 10-2 VFs from 289 eyes (257 patients). Macular scans were used to estimate 10-2 VF mean deviation (MD), threshold sensitivities (TS), and total deviation (TD) values at 68 locations. A three-dimensional (3D) convolutional neural network based on the 3D DenseNet121 architecture was used for prediction. We compared DL predictions to those from baseline linear models. We carried out 10-fold stratified cross-validation to optimize generalizability. The performance of the DL and baseline models was compared based on correlations between ground truth and predicted VF measures and mean absolute error (MAE; ground truth - predicted values). Results: Average (SD) MD was -9.3 (7.7) dB. Average (SD) correlations between predicted and ground truth MD and MD MAE were 0.74 (0.09) and 3.5 (0.4) dB, respectively. Estimation accuracy deteriorated with worsening MD. Average (SD) Pearson correlations between predicted and ground truth TS and MAEs for DL and baseline model were 0.71 (0.05) and 0.52 (0.05) (P < 0.001) and 6.5 (0.6) and 7.5 (0.5) dB (P < 0.001), respectively. For TD, correlation (SD) and MAE (SD) for DL and baseline models were 0.69 (0.02) and 0.48 (0.05) (P < 0.001) and 6.1 (0.5) and 7.8 (0.5) dB (P < 0.001), respectively. Conclusions: Macular OCT volume scans can be used to predict global central VF parameters with clinically relevant accuracy. Translational Relevance: Macular OCT imaging may be used to confirm and supplement central VF findings using deep learning.

Prediction of visual field progression with serial optic disc photographs using deep learning.

AIM: We tested the hypothesis that visual field (VF) progression can be predicted with a deep learning model based on longitudinal pairs of optic disc photographs (ODP) acquired at earlier time points during follow-up. METHODS: 3919 eyes (2259 patients) with ≥2 ODPs at least 2 years apart, and ≥5 24-2 VF exams spanning ≥3 years of follow-up were included. Serial VF mean deviation (MD) rates of change were estimated starting at the fifth visit and subsequently by adding visits until final visit. VF progression was defined as a statistically significant negative slope at two consecutive visits and final visit. We built a twin-neural network with ResNet50-backbone. A pair of ODPs acquired up to a year before the VF progression date or the last VF in non-progressing eyes were included as input. Primary outcome measures were area under the receiver operating characteristic curve (AUC) and model accuracy. RESULTS: The average (SD) follow-up time and baseline VF MD were 8.1 (4.8) years and -3.3 (4.9) dB, respectively. VF progression was identified in 761 eyes (19%). The median (IQR) time to progression in progressing eyes was 7.3 (4.5-11.1) years. The AUC and accuracy for predicting VF progression were 0.862 (0.812-0.913) and 80.0% (73.9%-84.6%). When only fast-progressing eyes were considered (MD rate < -1.0 dB/year), AUC increased to 0.926 (0.857-0.994). CONCLUSIONS: A deep learning model can predict subsequent glaucoma progression from longitudinal ODPs with clinically relevant accuracy. This model may be implemented, after validation, for predicting glaucoma progression in the clinical setting.

Mask R-CNN assisted 2.5D object detection pipeline of 68Ga-PSMA-11 PET/CT-positive metastatic pelvic lymph node after radical prostatectomy from solely CT imaging.

Prostate-specific membrane antigen (PSMA) positron emission tomography (PET)/computed tomography (CT) is a molecular and functional imaging modality with better restaging accuracy over conventional imaging for detecting prostate cancer in men suspected of lymph node (LN) progression after definitive therapy. However, the availability of PSMA PET/CT is limited in both low-resource settings and for repeating imaging surveillance. In contrast, CT is widely available, cost-effective, and routinely performed as part of patient follow-up or radiotherapy workflow. Compared with the molecular activities, the morphological and texture changes of subclinical LNs in CT are subtle, making manual detection of positive LNs infeasible. Instead, we harness the power of artificial intelligence for automated LN detection on CT. We examined 68Ga-PSMA-11 PET/CT images from 88 patients (including 739 PSMA PET/CT-positive pelvic LNs) who experienced a biochemical recurrence after radical prostatectomy and presented for salvage radiotherapy with prostate-specific antigen < 1 ng/mL. Scans were divided into a training set (nPatient = 52, nNode = 400), a validation set (nPatient = 18, nNode = 143), and a test set (nPatient = 18, nNodes = 196). Using PSMA PET/CT as the ground truth and consensus pelvic LN clinical target volumes as search regions, a 2.5-dimensional (2.5D) Mask R-CNN based object detection framework was trained. The entire framework contained whole slice imaging pretraining, masked-out region fine-tuning, prediction post-processing, and "window bagging". Following an additional preprocessing step-pelvic LN clinical target volume extraction, our pipeline located positive pelvic LNs solely based on CT scans. Our pipeline could achieve a sensitivity of 83.351%, specificity of 58.621% out of 196 positive pelvic LNs from 18 patients in the test set, of which most of the false positives can be post-removable by radiologists. Our tool may aid CT-based detection of pelvic LN metastasis and triage patients most unlikely to benefit from the PSMA PET/CT scan.

Predicting Hypoperfusion Lesion and Target Mismatch in Stroke from Diffusion-weighted MRI Using Deep Learning.

Background Perfusion imaging is important to identify a target mismatch in stroke but requires contrast agents and postprocessing software. Purpose To use a deep learning model to predict the hypoperfusion lesion in stroke and identify patients with a target mismatch profile from diffusion-weighted imaging (DWI) and clinical information alone, using perfusion MRI as the reference standard. Materials and Methods Imaging data sets of patients with acute ischemic stroke with baseline perfusion MRI and DWI were retrospectively reviewed from multicenter data available from 2008 to 2019 (Imaging Collaterals in Acute Stroke, Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution 2, and University of California, Los Angeles stroke registry). For perfusion MRI, rapid processing of perfusion and diffusion software automatically segmented the hypoperfusion lesion (time to maximum, ≥6 seconds) and ischemic core (apparent diffusion coefficient [ADC], ≤620 × 10-6 mm2/sec). A three-dimensional U-Net deep learning model was trained using baseline DWI, ADC, National Institutes of Health Stroke Scale score, and stroke symptom sidedness as inputs, with the union of hypoperfusion and ischemic core segmentation serving as the ground truth. Model performance was evaluated using the Dice score coefficient (DSC). Target mismatch classification based on the model was compared with that of the clinical-DWI mismatch approach defined by the DAWN trial by using the McNemar test. Results Overall, 413 patients (mean age, 67 years ± 15 [SD]; 207 men) were included for model development and primary analysis using fivefold cross-validation (247, 83, and 83 patients in the training, validation, and test sets, respectively, for each fold). The model predicted the hypoperfusion lesion with a median DSC of 0.61 (IQR, 0.45-0.71). The model identified patients with target mismatch with a sensitivity of 90% (254 of 283; 95% CI: 86, 93) and specificity of 77% (100 of 130; 95% CI: 69, 83) compared with the clinical-DWI mismatch sensitivity of 50% (140 of 281; 95% CI: 44, 56) and specificity of 89% (116 of 130; 95% CI: 83, 94) (P < .001 for all). Conclusion A three-dimensional U-Net deep learning model predicted the hypoperfusion lesion from diffusion-weighted imaging (DWI) and clinical information and identified patients with a target mismatch profile with higher sensitivity than the clinical-DWI mismatch approach. ClinicalTrials.gov registration nos. NCT02225730, NCT01349946, NCT02586415 © RSNA, 2022 Supplemental material is available for this article. See also the editorial by Kallmes and Rabinstein in this issue.

Machine learning approach for obstructive sleep apnea screening using brain diffusion tensor imaging.

Patients with obstructive sleep apnea (OSA) show autonomic, mood, cognitive, and breathing dysfunctions that are linked to increased morbidity and mortality, which can be improved with early screening and intervention. The gold standard and other available methods for OSA diagnosis are complex, require whole-night data, and have significant wait periods that potentially delay intervention. Our aim was to examine whether using faster and less complicated machine learning models, including support vector machine (SVM) and random forest (RF), with brain diffusion tensor imaging (DTI) data can classify OSA from healthy controls. We collected two DTI series from 59 patients with OSA [age: 50.2 ± 9.9 years; body mass index (BMI): 31.5 ± 5.6 kg/m2 ; apnea-hypopnea index (AHI): 34.1 ± 21.2 events/h 23 female] and 96 controls (age: 51.8 ± 9.7 years; BMI: 26.2 ± 4.1 kg/m2 ; 51 female) using a 3.0-T magnetic resonance imaging scanner. Using DTI data, mean diffusivity maps were calculated from each series, realigned and averaged, normalised to a common space, and used to conduct cross-validation for model training and selection and to predict OSA. The RF model showed 0.73 OSA and controls classification accuracy and 0.85 area under the curve (AUC) value on the receiver-operator curve. Cross-validation showed the RF model with comparable fitting over SVM for OSA and control data (SVM; accuracy, 0.77; AUC, 0.84). The RF ML model performs similar to SVM, indicating the comparable statistical fitness to DTI data. The findings indicate that RF model has similar AUC and accuracy over SVM, and either model can be used as a faster OSA screening tool for subjects having brain DTI data.

Fully automated segmentation of prostatic urethra for MR-guided radiation therapy.

PURPOSE: Accurate delineation of the urethra is a prerequisite for urethral dose reduction in prostate radiotherapy. However, even in magnetic resonance-guided radiation therapy (MRgRT), consistent delineation of the urethra is challenging, particularly in online adaptive radiotherapy. This paper presented a fully automatic MRgRT-based prostatic urethra segmentation framework. METHODS: Twenty-eight prostate cancer patients were included in this study. In-house 3D half fourier single-shot turbo spin-echo (HASTE) and turbo spin echo (TSE) sequences were used to image the Foley-free urethra on a 0.35 T MRgRT system. The segmentation pipeline uses 3D nnU-Net as the base and innovatively combines ground truth and its corresponding radial distance (RD) map during training supervision. Additionally, we evaluate the benefit of incorporating a convolutional long short term memory (LSTM-Conv) layer and spatial recurrent convolution layer (RCL) into nnU-Net. A novel slice-by-slice simple exponential smoothing (SEPS) method specifically for tubular structures was used to post-process the segmentation results. RESULTS: The experimental results show that nnU-Net trained using a combination of Dice, cross-entropy and RD achieved a Dice score of 77.1 ± 2.3% in the testing dataset. With SEPS, Hausdorff distance (HD) and 95% HD were reduced to 2.95 ± 0.17 mm and 1.84 ± 0.11 mm, respectively. LSTM-Conv and RCL layers only minimally improved the segmentation precision. CONCLUSION: We present the first Foley-free MRgRT-based automated urethra segmentation study. Our method is built on a data-driven neural network with novel cost functions and a post-processing step designed for tubular structures. The performance is consistent with the need for online and offline urethra dose reduction in prostate radiotherapy.

A mobile battery-powered brain perfusion ultrasound (BPU) device designed for prehospital stroke diagnosis: correlation to perfusion MRI in healthy volunteers.

BACKGROUND: Early prehospital stroke identification is crucial for goal directed hospital admission especially in rural areas. However, clinical prehospital stroke scales are designed to identify any stroke but cannot sufficiently differentiate hemorrhagic from ischemic stroke, including large vessel occlusion (LVO) amenable to mechanical thrombectomy. We report on a novel small, portable and battery driven point-of-care ultrasound system (SONAS®) specifically developed for mobile non-invasive brain perfusion ultrasound (BPU) measurement after bolus injection of an echo-enhancing agent suitable for the use in prehospital stroke diagnosis filling a current, unmet and critical need for LVO identification. METHODS: In a phase I study of healthy volunteers we performed comparative perfusion-weighted magnetic resonance imaging (PWI) and BPU measurements, including safety analysis. RESULTS: Twelve volunteers (n = 7 females, n = 5 males, age ranging between 19 and 55 years) tolerated the measurement extremely well including analysis of blood-brain barrier integrity, and the correlation coefficient between the generated time kinetic curves after contrast agent bolus between PWI and BPU transducers ranged between 0.89 and 0.76. CONCLUSIONS: Mobile BPU using the SONAS® device is feasible and safe with results comparable to PWI. When applied in conjunction with prehospital stroke scales this may lead to a more accurate stroke diagnosis and patients bypassing regular stroke units to comprehensive stroke centers. Further studies are needed in acute stroke patients and in the prehospital phase including assessment of immediate and long-term morbidity and mortality in stroke. TRIAL REGISTRATION: Clinical trials.gov, registered 28.Sep.2017, Identifier: NCT03296852.

Quantification of infarct core signal using CT imaging in acute ischemic stroke.

In stroke care, the extent of irreversible brain injury, termed infarct core, plays a key role in determining eligibility for acute treatments, such as intravenous thrombolysis and endovascular reperfusion therapies. Many of the pivotal randomized clinical trials testing those therapies used MRI Diffusion-Weighted Imaging (DWI) or CT Perfusion (CTP) to define infarct core. Unfortunately, these modalities are not available 24/7 outside of large stroke centers. As such, there is a need for accurate infarct core determination using faster and more widely available imaging modalities including Non-Contrast CT (NCCT) and CT Angiography (CTA). Prior studies have suggested that CTA provides improved predictions of infarct core relative to NCCT; however, this assertion has never been numerically quantified by automatic medical image computing pipelines using acquisition protocols not confounded by different scanner manufacturers, or other protocol settings such as exposure times, kilovoltage peak, or imprecision due to contrast bolus delays. In addition, single-phase CTA protocols are at present designed to optimize contrast opacification in the arterial phase. This approach works well to maximize the sensitivity to detect vessel occlusions, however, it may not be the ideal timing to enhance the ischemic infarct core signal (ICS). In this work, we propose an image analysis pipeline on CT-based images of 88 acute ischemic stroke (AIS) patients drawn from a single dynamic acquisition protocol acquired at the acute ischemic phase. We use the first scan at the time of the dynamic acquisition as a proxy for NCCT, and the rest of the scans as a proxy for CTA scans, with bolus imaged at different brain enhancement phases. Thus, we use the terms "NCCT" and "CTA" to refer to them. This pipeline enables us to answer the questions "Does the injection of bolus enhance the infarct core signal?" and "What is the ideal bolus timing to enhance the infarct core signal?" without being influenced by aforementioned factors such as scanner model, acquisition settings, contrast bolus delay, and human reader errors. We use reference MRI DWI images acquired after successful recanalization acting as our gold standard for infarct core. The ICS is quantified by calculating the difference in intensity distribution between the infarct core region and its symmetrical healthy counterpart on the contralateral hemisphere of the brain using a metric derived from information theory, the Kullback-Leibler divergence (KL divergence). We compare the ICS provided by NCCT and CTA and retrieve the optimal timing of CTA bolus to maximize the ICS. In our experiments, we numerically confirm that CTAs provide greater ICS compared to NCCT. Then, we find that, on average, the ideal CTA acquisition time to maximize the ICS is not the current target of standard CTA protocols, i.e., during the peak of arterial enhancement, but a few seconds afterward (median of 3 s; 95% CI [1.5, 3.0]). While there are other studies comparing the prediction potential of ischemic infarct core from NCCT and CTA images, to the best of our knowledge, this analysis is the first to perform a quantitative comparison of the ICS among CT based scans, with and without bolus injection, acquired using the same scanning sequence and a precise characterization of the bolus uptake, hence, reducing potential confounding factors.

Multiparametric MRI-based radiomics model to predict pelvic lymph node invasion for patients with prostate cancer.

OBJECTIVE: To identify which patient with prostate cancer (PCa) could safely avoid extended pelvic lymph node dissection (ePLND) by predicting lymph node invasion (LNI), via a radiomics-based machine learning approach. METHODS: An integrative radiomics model (IRM) was proposed to predict LNI, confirmed by the histopathologic examination, integrating radiomics features, extracted from prostatic index lesion regions on MRI images, and clinical features via SVM. The study cohort comprised 244 PCa patients with MRI and followed by radical prostatectomy (RP) and ePLND within 6 months between 2010 and 2019. The proposed IRM was trained in training/validation set and evaluated in an internal independent testing set. The model's performance was measured by area under the curve (AUC), sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV). AUCs were compared via Delong test with 95% confidence interval (CI), and the rest measurements were compared via chi-squared test or Fisher's exact test. RESULTS: Overall, 17 (10.6%) and 14 (16.7%) patients with LNI were included in training/validation set and testing set, respectively. Shape and first-order radiomics features showed usefulness in building the IRM. The proposed IRM achieved an AUC of 0.915 (95% CI: 0.846-0.984) in the testing set, superior to pre-existing nomograms whose AUCs were from 0.698 to 0.724 (p < 0.05). CONCLUSION: The proposed IRM could be potentially feasible to predict the risk of having LNI for patients with PCa. With the improved predictability, it could be utilized to assess which patients with PCa could safely avoid ePLND, thus reduce the number of unnecessary ePLND. KEY POINTS: • The combination of MRI-based radiomics features with clinical information improved the prediction of lymph node invasion, compared with the model using only radiomics features or clinical features. • With improved prediction performance on predicting lymph node invasion, the number of extended pelvic lymph node dissection (ePLND) could be reduced by the proposed integrative radiomics model (IRM), compared with the existing nomograms.

Determining and Validating Population Differences in Magnetic Resonance Angiography Using Sparse Representation.

Goal: Identifying population differences can serve as an insightful tool for diagnostic radiology. To do so, a reliable preprocessing framework and data representation are vital. Methods: We build a machine learning model to visualize gender differences in the circle of Willis (CoW), an integral part of the brain's vasculature. We start with a dataset of 570 individuals and process them for analysis using 389 for the final analysis. Results: We find statistical differences between male and female patients in one image plane and visualize where they are. We can see differences between the right and left-hand sides of the brain confirmed using Support Vector Machines (SVM). Conclusion: This process can be applied to detect population variations in the vasculature automatically. Significance: It can guide debugging and inferring complex machine learning algorithms such as SVM and deep learning models.

Heterogeneity between proximal and distal aspects of occlusive thrombi on pretreatment imaging in acute ischemic stroke.

BACKGROUND: The potential heterogeneity in occlusive thrombi caused by in situ propagation by secondary thrombosis after embolic occlusion could obscure the characteristics of original thrombi, preventing the clarification of a specific thrombus signature for the etiology of ischemic stroke. We aimed to investigate the heterogeneity of occlusive thrombi by pretreatment imaging. METHODS: Among consecutive stroke patients with acute embolic anterior circulation large vessel occlusion treated with thrombectomy, we retrospectively reviewed 104 patients with visible occlusive thrombi on pretreatment non-contrast computed tomography admitted from January 2015 to December 2018. A region of interest was set on the whole thrombus on non-contrast computed tomography under the guidance of computed tomography angiography. The region of interest was divided equally into the proximal and distal segments and the difference in Hounsfield unit densities between the two segments was calculated. RESULTS: Hounsfield unit density in the proximal segment was higher than that in the distal segment (mean difference 4.45; p < 0.001), regardless of stroke subtypes. On multivariate analysis, thrombus length was positively correlated (ß = 0.25; p < 0.001) and time from last-known-well to imaging was inversely correlated (ß = -0.0041; p = 0.002) with the difference in Hounsfield unit densities between the proximal and distal segments. CONCLUSIONS: The difference in density between the proximal and distal segments increased as thrombi became longer and decreased as thrombi became older after embolic occlusion. This time/length-dependent thrombus heterogeneity between the two segments is suggestive of secondary thrombosis initially occurring on the proximal side of the occlusion.

Integrative Machine Learning Prediction of Prostate Biopsy Results From Negative Multiparametric MRI.

BACKGROUND: Multiparametric MRI (mpMRI) is commonly recommended as a triage test prior to any prostate biopsy. However, there exists limited consensus on which patients with a negative prostate mpMRI could avoid prostate biopsy. PURPOSE: To identify which patient could safely avoid prostate biopsy when the prostate mpMRI is negative, via a radiomics-based machine learning approach. STUDY TYPE: Retrospective. SUBJECTS: Three hundred thirty patients with negative prostate 3T mpMRI between January 2016 and December 2018 were included. FIELD STRENGTH/SEQUENCE: A 3.0 T/T2-weighted turbo spin echo (TSE) imaging (T2 WI) and diffusion-weighted imaging (DWI). ASSESSMENT: The integrative machine learning (iML) model was trained to predict negative prostate biopsy results, utilizing both radiomics and clinical features. The final study cohort comprised 330 consecutive patients with negative mpMRI (PI-RADS < 3) who underwent systematic transrectal ultrasound-guided (TRUS) or MR-ultrasound fusion (MRUS) biopsy within 6 months. A secondary analysis of biopsy naïve subcohort (n = 227) was also conducted. STATISTICAL TESTS: The Mann-Whitney U test and Chi-Squared test were utilized to evaluate the significance of difference of clinical features between prostate biopsy positive and negative groups. The model performance was validated using leave-one-out cross-validation (LOOCV) and measured by AUC, sensitivity, specificity, and negative predictive value (NPV). RESULTS: Overall, 306/330 (NPV 92.7%) of the final study cohort patients had negative biopsies, and 207/227 (NPV 91.2%) of the biopsy naïve subcohort patients had negative biopsies. Our iML model achieved NPVs of 98.3% and 98.0% for the study cohort and subcohort, respectively, superior to prostate-specific antigen density (PSAD)-based risk assessment with NPVs of 94.9% and 93.9%, respectively. DATA CONCLUSION: The proposed iML model achieved high performance in predicting negative prostate biopsy results for patients with negative mpMRI. With improved NPVs, the proposed model can be used to stratify patients who in whom we might obviate biopsies, thus reducing the number of unnecessary biopsies. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Intra-domain task-adaptive transfer learning to determine acute ischemic stroke onset time.

Treatment of acute ischemic strokes (AIS) is largely contingent upon the time since stroke onset (TSS). However, TSS may not be readily available in up to 25% of patients with unwitnessed AIS. Current clinical guidelines for patients with unknown TSS recommend the use of MRI to determine eligibility for thrombolysis, but radiology assessments have high inter-reader variability. In this work, we present deep learning models that leverage MRI diffusion series to classify TSS based on clinically validated thresholds. We propose an intra-domain task-adaptive transfer learning method, which involves training a model on an easier clinical task (stroke detection) and then refining the model with different binary thresholds of TSS. We apply this approach to both 2D and 3D CNN architectures with our top model achieving an ROC-AUC value of 0.74, with a sensitivity of 0.70 and a specificity of 0.81 for classifying TSS < 4.5 h. Our pretrained models achieve better classification metrics than the models trained from scratch, and these metrics exceed those of previously published models applied to our dataset. Furthermore, our pipeline accommodates a more inclusive patient cohort than previous work, as we did not exclude imaging studies based on clinical, demographic, or image processing criteria. When applied to this broad spectrum of patients, our deep learning model achieves an overall accuracy of 75.78% when classifying TSS < 4.5 h, carrying potential therapeutic implications for patients with unknown TSS.

Intracranial atherosclerotic disease mechanistic subtypes drive hypoperfusion patterns.

BACKGROUND AND PURPOSE: In symptomatic intracranial atherosclerotic stenosis (ICAS), borderzone infarct pattern and perfusion mismatch are associated with increased risk of recurrent strokes, which may reflect the shared underlying mechanism of hypoperfusion distal to the intracranial atherosclerosis. Accordingly, we hypothesized a correlation between hypoperfusion volumes and ICAS infarct patterns based on the respective underlying mechanistic subtypes. METHODS: We conducted a retrospective analysis of consecutive symptomatic ICAS cases, acute strokes due to subocclusive (50%-99%) intracranial stenosis. The following mechanistic subtypes were assigned based on the infarct pattern on the diffusion-weighted imaging: Branch occlusive disease (BOD), internal borderzone (IBZ), and thromboembolic (TE). Perfusion parameters, obtained concurrently with the MRI, were studied in each group. RESULTS: A total of 42 patients (57% women, mean age 71 ± 13 years old) with symptomatic ICAS received MRI within 24 h of acute presentation. Fourteen IBZ, 11 BOD, and 17 TE patterns were identified. IBZ pattern yielded higher total Tmax > 4 s and Tmax > 6 s perfusion delay volumes, as well as corresponding Tmax  > 4 s and Tmax  > 6 s mismatch volume, compared to BOD. TE pattern exhibited greater median Tmax  > 6 s hypoperfusion delay in volume compared to BOD. In IBZ versus TE, the volume difference between Tmax > 4 s and Tmax > 6 s (Δ Tmax  > 4 s - Tmax  > 6 s) was substantially greater. CONCLUSION: ICAS infarct patterns, in keeping with their respective underlying mechanisms, may correlate with distinct perfusion profiles.

Performance of Deep Learning and Genitourinary Radiologists in Detection of Prostate Cancer Using 3-T Multiparametric Magnetic Resonance Imaging.

BACKGROUND: Several deep learning-based techniques have been developed for prostate cancer (PCa) detection using multiparametric magnetic resonance imaging (mpMRI), but few of them have been rigorously evaluated relative to radiologists' performance or whole-mount histopathology (WMHP). PURPOSE: To compare the performance of a previously proposed deep learning algorithm, FocalNet, and expert radiologists in the detection of PCa on mpMRI with WMHP as the reference. STUDY TYPE: Retrospective, single-center study. SUBJECTS: A total of 553 patients (development cohort: 427 patients; evaluation cohort: 126 patients) who underwent 3-T mpMRI prior to radical prostatectomy from October 2010 to February 2018. FIELD STRENGTH/SEQUENCE: 3-T, T2-weighted imaging and diffusion-weighted imaging. ASSESSMENT: FocalNet was trained on the development cohort to predict PCa locations by detection points, with a confidence value for each point, on the evaluation cohort. Four fellowship-trained genitourinary (GU) radiologists independently evaluated the evaluation cohort to detect suspicious PCa foci, annotate detection point locations, and assign a five-point suspicion score (1: least suspicious, 5: most suspicious) for each annotated detection point. The PCa detection performance of FocalNet and radiologists were evaluated by the lesion detection sensitivity vs. the number of false-positive detections at different thresholds on suspicion scores. Clinically significant lesions: Gleason Group (GG) ≥ 2 or pathological size ≥ 10 mm. Index lesions: the highest GG and the largest pathological size (secondary). STATISTICAL TESTS: Bootstrap hypothesis test for the detection sensitivity between radiologists and FocalNet. RESULTS: For the overall differential detection sensitivity, FocalNet was 5.1% and 4.7% below the radiologists for clinically significant and index lesions, respectively; however, the differences were not statistically significant (P = 0.413 and P = 0.282, respectively). DATA CONCLUSION: FocalNet achieved slightly lower but not statistically significant PCa detection performance compared with GU radiologists. Compared with radiologists, FocalNet demonstrated similar detection performance for a highly sensitive setting (suspicion score ≥ 1) or a highly specific setting (suspicion score = 5), while lower performance in between. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 2.