Optimizing double-layered convolutional neural networks for efficient lung cancer classification through hyperparameter optimization and advanced image pre-processing techniques.

Lung cancer remains a leading cause of cancer-related mortality globally, with prognosis significantly dependent on early-stage detection. Traditional diagnostic methods, though effective, often face challenges regarding accuracy, early detection, and scalability, being invasive, time-consuming, and prone to ambiguous interpretations. This study proposes an advanced machine learning model designed to enhance lung cancer stage classification using CT scan images, aiming to overcome these limitations by offering a faster, non-invasive, and reliable diagnostic tool. Utilizing the IQ-OTHNCCD lung cancer dataset, comprising CT scans from various stages of lung cancer and healthy individuals, we performed extensive preprocessing including resizing, normalization, and Gaussian blurring. A Convolutional Neural Network (CNN) was then trained on this preprocessed data, and class imbalance was addressed using Synthetic Minority Over-sampling Technique (SMOTE). The model's performance was evaluated through metrics such as accuracy, precision, recall, F1-score, and ROC curve analysis. The results demonstrated a classification accuracy of 99.64%, with precision, recall, and F1-score values exceeding 98% across all categories. SMOTE significantly enhanced the model's ability to classify underrepresented classes, contributing to the robustness of the diagnostic tool. These findings underscore the potential of machine learning in transforming lung cancer diagnostics, providing high accuracy in stage classification, which could facilitate early detection and tailored treatment strategies, ultimately improving patient outcomes.

Dimension reduction and outlier detection of 3-D shapes derived from multi-organ CT images.

BACKGROUND: Unsupervised clustering and outlier detection are important in medical research to understand the distributional composition of a collective of patients. A number of clustering methods exist, also for high-dimensional data after dimension reduction. Clustering and outlier detection may, however, become less robust or contradictory if multiple high-dimensional data sets per patient exist. Such a scenario is given when the focus is on 3-D data of multiple organs per patient, and a high-dimensional feature matrix per organ is extracted. METHODS: We use principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE) and multiple co-inertia analysis (MCIA) combined with bagplots to study the distribution of multi-organ 3-D data taken by computed tomography scans. After point-set registration of multiple organs from two public data sets, multiple hundred shape features are extracted per organ. While PCA and t-SNE can only be applied to each organ individually, MCIA can project the data of all organs into the same low-dimensional space. RESULTS: MCIA is the only approach, here, with which data of all organs can be projected into the same low-dimensional space. We studied how frequently (i.e., by how many organs) a patient was classified to belong to the inner or outer 50% of the population, or as an outlier. Outliers could only be detected with MCIA and PCA. MCIA and t-SNE were more robust in judging the distributional location of a patient in contrast to PCA. CONCLUSIONS: MCIA is more appropriate and robust in judging the distributional location of a patient in the case of multiple high-dimensional data sets per patient. It is still recommendable to apply PCA or t-SNE in parallel to MCIA to study the location of individual organs.

Dosimetric impact of contour editing on CT and MRI deep-learning autosegmentation for brain OARs.

PURPOSE: To establish the clinical applicability of deep-learning organ-at-risk autocontouring models (DL-AC) for brain radiotherapy. The dosimetric impact of contour editing, prior to model training, on performance was evaluated for both CT and MRI-based models. The correlation between geometric and dosimetric measures was also investigated to establish whether dosimetric assessment is required for clinical validation. METHOD: CT and MRI-based deep learning autosegmentation models were trained using edited and unedited clinical contours. Autosegmentations were dosimetrically compared to gold standard contours for a test cohort. D1%, D5%, D50%, and maximum dose were used as clinically relevant dosimetric measures. The statistical significance of dosimetric differences between the gold standard and autocontours was established using paired Student's t-tests. Clinically significant cases were identified via dosimetric headroom to the OAR tolerance. Pearson's Correlations were used to investigate the relationship between geometric measures and absolute percentage dose changes for each autosegmentation model. RESULTS: Except for the right orbit, when delineated using MRI models, the dosimetric statistical analysis revealed no superior model in terms of the dosimetric accuracy between the CT DL-AC models or between the MRI DL-AC for any investigated brain OARs. The number of patients where the clinical significance threshold was exceeded was higher for the optic chiasm D1% than other OARs, for all autosegmentation models. A weak correlation was consistently observed between the outcomes of dosimetric and geometric evaluations. CONCLUSIONS: Editing contours before training the DL-AC model had no significant impact on dosimetry. The geometric test metrics were inadequate to estimate the impact of contour inaccuracies on dose. Accordingly, dosimetric analysis is needed to evaluate the clinical applicability of DL-AC models in the brain.

Radiological examination of greater palatine canal medial wall dehiscence.

INTRODUCTION: The greater palatine canal (GPC) connects the pterygopalatine fossa to the greater palatine foramen and houses vital neurovascular structures, which provide sensory innervation and circulation to the gums, palate, and nasal cavity. The GPC is of great clinical importance to various medical specialties; however, the anatomical variability of the GPC poses a risk of iatrogenic injury and complications. Therefore, understanding the normal anatomy and variations of the GPC is crucial for identifying vital structures and minimizing risks in clinical practice. PURPOSE: The aim was to fill a gap in the current literature by focusing on the prevalence of GPC medial wall dehiscence, a lesser-known anatomic variation, in radiological scans. METHODS: A total of 200 head and neck CT scans were examined, where 71 scans met the inclusion criteria. Statistical significance for incidence of GPC medial wall dehiscence, in reference to sex and side, was measured. RESULTS: The GPC medial wall dehiscence was observed in 69% of scans. Bilateral dehiscence was seen in 57.7% of scans, while right-sided and left-sided unilateral dehiscence were found in 14.1% and 11.3%, respectively. Significant difference was found between the incidence of bilateral dehiscence compared to the absence of dehiscence. CONCLUSION: Previous studies have highlighted the potential risks associated with invasive procedures involving the GPC. The clinical relevance of GPC medial wall dehiscence lies in the increased risk of transecting the contained neurovascular bundle. The presence of dehiscence emphasizes the need for meticulous preoperative radiologic analysis to tailor surgical approaches to individual patient anatomy.

Indian Hand Anthropometry: Computer Tomography-Based Morphometric Analysis of Metacarpal.

Introduction Metacarpal fractures are common and have various treatment options, but understanding their morphometry is crucial for optimizing fixation techniques and reducing complications. Accurate assessment of metacarpal anatomy is challenging in conventional radiographs but feasible with computed tomography (CT) scans, which offer precise views. This study aimed to provide accurate anatomical data on metacarpals within an Indian population using CT scans and to compare the results with existing literature. The findings have implications for surgical procedures, including plating, pinning, and intramedullary screw fixation. Materials and Methods This retrospective analysis utilized CT scans of 100 hands, including 50 males and 50 females, from two hospitals in India. Inclusion criteria included complete metacarpal visualization with a slice thickness of 0.6 mm, while exclusion criteria involved trauma, deformity, or underlying pathologies. Various parameters of all metacarpals were measured using RadiAnt DICOM Viewer 2021.1, providing accurate anteroposterior and lateral views. Results Male and female cohorts had mean ages of 38.58 ± 12.02 and 43.60 ± 13.61 years, respectively. The study showed good to excellent reliability in measurements. The 2nd metacarpal was consistently the longest, and the general length pattern was 3rd > 4th > 5th > 1st metacarpal in both genders. Men generally had larger metacarpal dimensions than women, except for intramedullary diameter, which showed minimal sex-related differences. Notably, the medullary cavity's narrowest part was at the 4th metacarpal, and the thumb had the widest intramedullary diameter. Conclusion This study provides valuable anatomical reference data for metacarpals in an Indian population, aiding in optimizing surgical techniques for metacarpal fractures. The 2nd metacarpal consistently stood out as the longest, and men generally had larger metacarpal dimensions than women. These insights into anatomical variations can inform clinical decisions and stimulate further research in this field. However, a larger and more diverse sample would enhance the study's representativeness.

Sequential low-dose CT thorax scans to determine invasive pulmonary fungal infection incidence after allogeneic hematopoietic cell transplantation.

Invasive fungal disease (IFD) during neutropenia goes along with a high mortality for patients after allogeneic hematopoietic cell transplantation (alloHCT). Low-dose computed tomography (CT) thorax shows good sensitivity for the diagnosis of IFD with low radiation exposure. The aim of our study was to evaluate sequential CT thorax scans at two time points as a new reliable method to detect IFD during neutropenia after alloHCT. We performed a retrospective single-center observational study in 265/354 screened patients admitted for alloHCT from June 2015 to August 2019. All were examined by a low-dose CT thorax scan at admission (CT t0) and after stable neutrophil recovery (CT t1) to determine the incidences of IFD. Furthermore, antifungal prophylaxis medications were recorded and cohorts were analyzed for statistical differences in IFD incidence using the sequential CT scans. In addition, IFD cases were classified according to EORTC 2008. At CT t0 in 9.6% of the patients, an IFD was detected and antifungal therapy initiated. The cumulative incidence of IFD in CT t1 in our department was 14%. The use of Aspergillus-effective prophylaxis through voriconazole or posaconazole decreased CT thorax t1 suggesting IFD is statistically significant compared to prophylaxis with fluconazole (5.6% asp-azol group vs 16.3% fluconazole group, p = 0.048). In 86%, CT t1 was negative for IFD. Low-dose sequential CT thorax scans are a valuable tool to detect pulmonary IFDs and guide antifungal prophylaxis and therapies. Furthermore, a negative CT t1 scan shows a benefit by allowing discontinuation of antifungal medication sparing patients from drug interactions and side effects.

Radiation exposure in pediatric patients with early onset scoliosis: A longitudinal single-center study.

BACKGROUND: There is no consensus regarding the acceptable level of medical radiation exposure in patients with early-onset scoliosis. This study aimed to quantify radiation exposure in these patients and investigate factors associated with high exposure. METHODS: Patients with early-onset scoliosis who received care for their spine deformity and other comorbidities in our institution were retrospectively reviewed. Cumulative radiation exposure and total number of imaging studies were recorded. Patients with ≥30 mSv exposure were classified as high exposure and analyzed to clarify factors associated with high exposure. RESULTS: Thirty-five patients were included for analysis. The etiology of scoliosis was idiopathic in 8 patients, congenital in 7, syndromic in 8, and neuromuscular in 12. Fifteen patients underwent 19 spinal surgeries. The types of operation performed were definitive fusion (n = 12), vertebrectomy for hemivertebra (n = 2), growing rod (n = 1), lengthening (n = 3), and revision/partial implant removal (n = 1). The mean cumulative radiation dose was 22.3 mSv (range, 2.5-94.5 mSv). Spine radiography and computed tomography combined accounted for 15.0 mSv (range, 2.4-52.5 mSv, 67.3% of the mean cumulative dose). The mean radiation dose was significantly higher in patients who underwent spinal surgery than in those who did not (31.2 mSv vs. 15.6 mSv). The high-exposure group comprised 10 patients (1 idiopathic, 1 congenital, 5 syndromic, and 3 neuromuscular scoliosis) and 8 underwent 11 spinal operations. Among 8 patients who underwent spinal surgery, the cumulative radiation dose for spine was ≥30 mSv and spine computed tomography was performed an average of 4.0 times. CONCLUSIONS: Nearly one-third of patients with early-onset scoliosis and half of patients who underwent spinal surgery had >30 mSv radiation exposure due to multiple computed tomography. Medical radiation exposure and associated cancer risk should be considered when treating these patients.

COVID-19 CT-images diagnosis and severity assessment using machine learning algorithm.

As a pandemic, the primary evaluation tool for coronavirus (COVID-19) still has serious flaws. To improve the existing situation, all facilities and tools available in this field should be used to combat the pandemic. Reverse transcription polymerase chain reaction is used to evaluate whether or not a person has this virus, but it cannot establish the severity of the illness. In this paper, we propose a simple, reliable, and automatic system to diagnose the severity of COVID-19 from the CT scans into three stages: mild, moderate, and severe, based on the simple segmentation method and three types of features extracted from the CT images, which are ratio of infection, statistical texture features (mean, standard deviation, skewness, and kurtosis), GLCM and GLRLM texture features. Four machine learning techniques (decision trees (DT), K-nearest neighbors (KNN), support vector machines (SVM), and Naïve Bayes) are used to classify scans. 1801 scans are divided into four stages based on the CT findings in the scans and the description file found with the datasets. Our proposed model divides into four steps: preprocessing, feature extraction, classification, and performance evaluation. Four machine learning algorithms are used in the classification step: SVM, KNN, DT, and Naive Bayes. By SVM method, the proposed model achieves 99.12%, 98.24%, 98.73%, and 99.9% accuracy for COVID-19 infection segmentation at the normal, mild, moderate, and severe stages, respectively. The area under the curve of the model is 0.99. Finally, our proposed model achieves better performance than state-of-art models. This will help the doctors know the stage of the infection and thus shorten the time and give the appropriate dose of treatment for this stage.

Exploring the relationship between soft and hard tissues: The example of vertebral arteries and transverse foramina.

Understanding how the brain is provided with glucose and oxygen is of particular interest in human evolutionary studies. In addition to the internal carotid arteries, vertebral arteries contribute significantly to the cerebral and cerebellar blood flow. The size of the transverse foramina has been suggested to represent a reliable proxy for assessing the size of the vertebral arteries in fossil specimens. To test this assumption, here, we statistically explore spatial relationships between the transverse foramina and the vertebral arteries in extant humans. Contrast computed tomography (CT) scans of the cervical regions of 16 living humans were collected. Cross-sectional areas of the right and left transverse foramina and the corresponding vertebral arteries were measured on each cervical vertebra from C1 to C6 within the same individuals. The cross-sectional areas of the foramina and corresponding arteries range between 13.40 and 71.25 mm2 and between 4.53 and 29.40 mm2 , respectively. The two variables are significantly correlated except in C1. Using regression analyses, we generate equations that can be subsequently used to estimate the size of the vertebral arteries in fossil specimens. By providing additional evidence of intra- and inter-individual size variation of the arteries and corresponding foramina in extant humans, our study introduces an essential database for a better understanding of the evolutionary story of soft tissues in the fossil record.

Comparability of skeletal fibulae surfaces generated by different source scanning (dual-energy CT scan vs. high resolution laser scanning) and 3D geometric morphometric validation.

This work aims to test accuracy and comparability of 3D models of human skeletal fibulae generated by clinical CT and laser scanner virtual acquisitions. Mesh topology, segmentation and smoothing protocols were tested to assess variation among meshes generated with different scanning methods and procedures, and to evaluate meshes-interchangeability in 3D geometric morphometric analysis. A sample of 13 left human fibulae were scanned separately with Revolution Discovery CT dual energy (0.625 mm resolution) and ARTEC Space Spider 3D structured light laser scanner (0.1 mm resolution). Different segmentation methods, including half-maximum height (HMH) and MIA-clustering protocols, were compared to their high-resolution standard generated with laser-scanner by calculating topological surface deviations. Different smoothing algorithms were also evaluated, such as Laplacian and Taubin smoothing. A total of 142 semilandmarks were used to capture the shape of both proximal and distal fibular epiphyses. After Generalized Procrustes superimposition, the Procrustes coordinates of the proximal and distal fibular epiphyses were used separately to assess variation due to scanning methods and the operator error. Smoothing algorithms at low iteration do not provide significant variation among reconstructions, but segmentation protocol may influence final mesh quality (0.09-0.24 mm). Mean deviation among CT-generated meshes that were segmented with MIA-clustering protocol, and laser scanner-generated ones, is optimal (0.42 mm, ranging 0.35-0.56 mm). Principal component analysis reveals that homologous samples scanned with the two methods cluster together for both the proximal and distal fibular epiphyses. Similarly, Procrustes ANOVA reveals no shape differences between scanning methods and replicates, and only 1.38-1.43% of shape variation is due to scanning device. Topological similarities support the comparability of CT- and laser scanner-generated meshes and validate its simultaneous use in shape analysis with potential clinical relevance. We precautionarily suggest that dedicated trials should be performed in each study when merging different data sources prior to analyses.

Pre-treatment CT-based radiomics nomogram for predicting microsatellite instability status in colorectal cancer.

OBJECTIVES: Stratification of microsatellite instability (MSI) status in patients with colorectal cancer (CRC) improves clinical decision-making for cancer treatment. The present study aimed to develop a radiomics nomogram to predict the pre-treatment MSI status in patients with CRC. METHODS: A total of 762 patients with CRC confirmed by surgical pathology and MSI status determined with polymerase chain reaction (PCR) method were retrospectively recruited between January 2013 and May 2019. Radiomics features were extracted from routine pre-treatment abdominal pelvic computed tomography (CT) scans acquired as part of the patients' clinical care. A radiomics nomogram was constructed using multivariate logistic regression. The performance of the nomogram was evaluated using discrimination, calibration, and decision curves. RESULTS: The radiomics nomogram incorporating radiomics signatures, tumor location, patient age, high-density lipoprotein expression, and platelet counts showed good discrimination between patients with non-MSI-H and MSI-H, with an area under the curve (AUC) of 0.74 [95% CI, 0.68-0.80] in the training cohort and 0.77 [95% CI, 0.68-0.85] in the validation cohort. Favorable clinical application was observed using decision curve analysis. The addition of pathological characteristics to the nomogram failed to show incremental prognostic value. CONCLUSIONS: We developed a radiomics nomogram incorporating radiomics signatures and clinical indicators, which could potentially be used to facilitate the individualized prediction of MSI status in patients with CRC. KEY POINTS: • There is an unmet need to non-invasively determine MSI status prior to treatment. However, the traditional radiological evaluation of CT is limited for evaluating MSI status. • Our non-invasive CT imaging-based radiomics method could efficiently distinguish patients with high MSI disease from those with low MSI disease. • Our radiomics approach demonstrated promising diagnostic efficiency for MSI status, similar to the commonly used IHC method.

The relationship between retinal structure and visual function in non-immuno-compromised people living with HIV without retinitis on antiretroviral therapy.

BACKGROUND: This study sought to establish the association between retinal morphology, visual function and linear parameters of cerebral atrophy in non-immunocompromised people living with HIV (NIPLHIV). METHODS: Sixty participants (30 NIPLHIV, 30 controls), aged 18-45 years, were sourced from an outpatient clinic in South Africa. NIPLHIV on antiretroviral therapy (ART) had elevated CD4 counts and low viral loads. Macula thickness and volume measurements were obtained using the Spectralis optical coherence tomographer. Contrast sensitivity (CS), colour vision and visual-evoked potentials (VEP) were also obtained. Linear parameters of cerebral atrophy (Sylvian fissure ratio, SFR) and bicaudate nucleus ratio (BCR) were all acquired from computed tomography (CT) scans. Associations between retinal thickness and volume and visual function were established by principal component factor analysis. RESULTS: CS scores were indirectly associated with the Inner Nuclear Layer (INL)-ETDRS thickness and volume subfields (co-efficient = -0.07; p = 0.02 and -0.11; p = 0.001), respectively. F100 total error scores (TES) were directly associated with the thicknesses of Ganglion Cell Layer-ETDRS subfields (co-efficient = 6.06; p = 0.04) but indirectly associated with INL-ETDRS subfields (co-efficient = -5.49; p = 0.04). F100-TES were indirectly associated with volumes of RNFL (Retinal Nerve Fibre Layer)-ETDRS subfields (co-efficient = -5.54; p = 0.02) and inner retina -ETDRS subfields (co-efficient = -6.70; p = 0.02). P100 latency was directly associated with RNFL-ETDRS subfield thickness (co-efficient = 2.90; p = 0.02) and volumes of outer retina subfields (co-efficient = 2.72; p = 0.04). CS scores were directly associated with SFR (co-efficient = -0.04; p = 0.01). F100-TES were directly associated with BCR (co-efficient = 0.003; p = 0.004) and SFR (co-efficient = 0.002; p = 0.02). P100 latency was indirectly associated with BCR (co-efficient = -0.001; p = 0.03). CONCLUSION: The recognition of associations may be the first step in the proposal to develop a framework for the surveillance of vision in patients with NIPLHIV. We recommend a study of the sample population to track the stability of these observations before general recommendations for clinical care.

How much acromial bone should be removed to decrease the critical shoulder angle? A 3D CT acromioplasty planning study.

BACKGROUND: A high critical shoulder angle (CSA) is associated with rotator cuff tear (RCT) and retear rate after repair. CSA reduced to less than 33° by acromioplasty is correlated with better clinical results and healing. But up to 24% of patients retain a CSA above 35° after acromioplasty. The objective of the study was to evaluate the use of 3D when planning acromioplasty and measure acromial bone removal volume dimensions. METHODS: Computed tomography (CT) scans from 45 patients with RCT and CSA ≥ 38° were retrospectively included. A 33° CSA cutting plane was positioned. Acromion was divided into 5 mm slices and acromial bone resection measured on each slice. RESULTS: Intra- and inter-observer reproducibility measurements were rated strong or very strong. Patients' mean preoperative CSA was 40° (38°, 49° ± 2.3°). Measurements of acromial resection were: anteroposterior length: 32.7 mm (20, 50 ± 7.4); inferior width: 7.6 mm (4.2, 19 ± 2.9); superior width: 4.1 mm (0, 16 ± 3.0); height: 6.1 mm (1.7, 6.7 ± 1.6); and cutting angle: 74° (46, 91 ± 8.0). Maximum width of acromial resection was located 10.6 mm (5, 17.5 ± 0.6) from the acromion's anterior edge and decreased gradually moving posteriorly. Preoperative CSA was linearly correlated with width (P < 0.0001, R = 84%) and length (P = 0.0001, R = 28%) of acromioplasty; the higher the CSA, the greater the width and length. CONCLUSIONS: 3D CT reconstructions are valid for planning a CSA decreasing acromioplasty. To reduce CSA to 33°, acromioplasty must be performed anterolaterally and resection is at least 2 cm long anteroposteriorly. For higher CSAs, acromioplasty may require lateral resection over 1 cm in width and up to 5 cm in length. To decrease the CSA efficiently, acromioplasty must be adapted to patient anatomy and 3D planning could be considered. LEVEL OF EVIDENCE: IV.

Multi-task multi-modality SVM for early COVID-19 Diagnosis using chest CT data.

In the early diagnosis of the Coronavirus disease (COVID-19), it is of great importance for either distinguishing severe cases from mild cases or predicting the conversion time that mild cases would possibly convert to severe cases. This study investigates both of them in a unified framework by exploring the problems such as slight appearance difference between mild cases and severe cases, the interpretability, the High Dimension and Low Sample Size (HDLSS) data, and the class imbalance. To this end, the proposed framework includes three steps: (1) feature extraction which first conducts the hierarchical segmentation on the chest Computed Tomography (CT) image data and then extracts multi-modality handcrafted features for each segment, aiming at capturing the slight appearance difference from different perspectives; (2) data augmentation which employs the over-sampling technique to augment the number of samples corresponding to the minority classes, aiming at investigating the class imbalance problem; and (3) joint construction of classification and regression by proposing a novel Multi-task Multi-modality Support Vector Machine (MM-SVM) method to solve the issue of the HDLSS data and achieve the interpretability. Experimental analysis on two synthetic and one real COVID-19 data set demonstrated that our proposed framework outperformed six state-of-the-art methods in terms of binary classification and regression performance.

BLCov: A novel collaborative-competitive broad learning system for COVID-19 detection from radiology images.

With the global outbreak of COVID-19, there is an urgent need to develop an effective and automated detection approach as a faster diagnostic alternative to avoid the spread of COVID-19. Recently, broad learning system (BLS) has been viewed as an alternative method of deep learning which has been applied to many areas. Nevertheless, the sparse autoencoder in classical BLS just considers the representations to reconstruct the input data but ignores the relationship among the extracted features. In this paper, inspired by the effectiveness of the collaborative-competitive representation (CCR) mechanism, a novel collaborative-competitive representation-based autoencoder (CCRAE) is first proposed, and then collaborative-competitive broad learning system (CCBLS) is proposed based on CCRAE to effectively address the issues mentioned above. Moreover, an automated CCBLS-based approach is proposed for COVID-19 detection from radiology images such as CT scans and chest X-ray images. In the proposed approach, a feature extraction module is utilized to extract features from CT scans or chest X-ray images, then we use these features for COVID-19 detection with CCBLS. The experimental results demonstrated that our proposed approach can achieve superior or comparable performance in comparison with ten other state-of-the-art methods.

COVID-CT-Mask-Net: prediction of COVID-19 from CT scans using regional features.

We present COVID-CT-Mask-Net model that predicts COVID-19 in chest CT scans. The model works in two stages: in the first stage, Mask R-CNN is trained to localize and detect two types of lesions in images. In the second stage, these detections are fused to classify the whole input image. To develop the solution for the three-class problem (COVID-19, Common Pneumonia and Control), we used the COVIDx-CT data split derived from the dataset of chest CT scans collected by China National Center for Bioinformation. We use 3000 images (about 5% of the train split of COVIDx-CT) to train the model. Without any complicated data normalization, balancing and regularization, and training only a small fraction of the model's parameters, we achieve a 9 0 . 8 0 % COVID-19 sensitivity, 9 1 . 6 2 % Common Pneumonia sensitivity and 9 2 . 1 0 % true negative rate (Control sensitivity), an overall accuracy of 9 1 . 6 6 % and F1-score of 9 1 . 5 0 % on the test data split with 21192 images, bringing the ratio of test to train data to 7.06. We also establish an important result that regional predictions (bounding boxes with confidence scores) detected by Mask R-CNN can be used to classify whole images. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

Pronation on weightbearing radiographs does not correlate with pronation from weightbearing CT scans.

INTRODUCTION: In hallux valgus (HV), first metatarsal pronation is increasingly recognized as an important aspect of the deformity. The purpose of this study was to compare pronation in HV patients determined from the shape of the lateral head of the first metatarsal on AP weightbearing radiographs with pronation calculated from weightbearing CT (WBCT) scans. METHODS: Patients were included in this study if they had preoperative and 5-month postoperative WBCT scans and corresponding weightbearing AP radiographs of the affected foot. Pronation of the first metatarsal on WBCT scans was measured using a 3D CAD model and the alpha angle and categorized into four groups on radiographs. Association between pronation groups on radiographs and WBCT scans was determined using Spearman correlation coefficients (rs) and by comparing mean WBCT pronation of the first metatarsal between plain radiograph pronation groups. RESULTS: Agreement between the two observers' pronation on radiographs was good (k = 0.634) and moderate (k = 0.501), respectively. There was no correlation between radiographic pronation and the 3D CAD model (rs < 0.15). Preoperatively, there was weak correlation between the alpha angle and the radiographic pronation groups (rs = 0.371, P = 0.048) although this relationship did not hold postoperatively (rs = 0.330, P = 0.081). There was no difference in mean pronation calculated on WBCT scans between the plain radiographic groups. CONCLUSION: Pronation of the first metatarsal measured on weightbearing AP radiographs had moderate interobserver agreement and was only weakly associated with pronation measured from WBCT scans. These results suggest that first metatarsal pronation measured on weightbearing radiographs is not a substitute for pronation measured on WBCT scans. LEVEL OF EVIDENCE: III, retrospective cohort study.

Associations between bone attenuation and prevalent vertebral fractures on chest CT scans differ with vertebral fracture locations.

Vertebral fracture (VF) locations are bimodally distributed in the spine. The association between VF and bone attenuation (BA) measured on chest CT scans varied according to the location of VFs, indicating that other factors than only BA play a role in the bimodal distribution of VFs. INTRODUCTION: Vertebral fractures (VFs) are associated with low bone mineral density but are not equally distributed throughout the spine and occur most commonly at T7-T8 and T11-T12 ("cVFs") and less commonly at T4-T6 and T9-T10 ("lcVF"). We aimed to determine whether associations between bone attenuation (BA) and VFs vary between subjects with cVFs only, with lcVFs only and with both cVFs and lcVFs. METHODS: Chest CT images of T4-T12 in 1237 smokers with and without COPD were analysed for prevalent VFs according to the method described by Genant (11,133 vertebrae). BA (expressed in Hounsfield units) was measured in all non-fractured vertebrae (available for 10,489 vertebrae). Linear regression was used to compare mean BA, and logistic regression was used to estimate the association of BA with prevalent VFs (adjusted for age and sex). RESULTS: On vertebral level, the proportion of cVFs was significantly higher than of lcVF (5.6% vs 2.0%). Compared to subjects without VFs, BA was 15% lower in subjects with cVFs (p < 0.0001), 25% lower in subjects with lcVFs (p < 0.0001) and lowest in subjects with cVFs and lcVFs (- 32%, p < 0.0001). The highest ORs for presence of VFs per - 1SD BA per vertebra were found in subjects with both cVFs and lcVFs (3.8 to 4.6). CONCLUSIONS: The association between VFs and BA differed according to VF location. ORs increased from subjects with cVFs to subjects with lcVFs and were highest in subjects with cVFs and lcVFs, indicating that other factors than only BA play a role in the bimodal VF distribution. TRIAL REGISTRATION: Clinicaltrials.gov identifier: NCT00292552.

Pre-surgical and Post-surgical Aortic Aneurysm Maximum Diameter Measurement: Full Automation by Artificial Intelligence.

OBJECTIVE: The aim of this study was to evaluate an automatic, deep learning based method (Augmented Radiology for Vascular Aneurysm [ARVA]), to detect and assess maximum aortic diameter, providing cross sectional outer to outer aortic wall measurements. METHODS: Accurate external aortic wall diameter measurement is performed along the entire aorta, from the ascending aorta to the iliac bifurcations, on both pre- and post-operative contrast enhanced computed tomography angiography (CTA) scans. A training database of 489 CTAs was used to train a pipeline of neural networks for automatic external aortic wall measurements. Another database of 62 CTAs, including controls, aneurysmal aortas, and aortic dissections scanned before and/or after endovascular or open repair, was used for validation. The measurements of maximum external aortic wall diameter made by ARVA were compared with those of seven clinicians on this validation dataset. RESULTS: The median absolute difference with respect to expert's measurements ranged from 1 mm to 2 mm among all annotators, while ARVA reported a median absolute difference of 1.2 mm. CONCLUSION: The performance of the automatic maximum aortic diameter method falls within the interannotator variability, making it a potentially reliable solution for assisting clinical practice.

Testing the reliability of CT scan-based dental wear magnitude scoring.

OBJECTIVES: Digital models are now frequently used in biological anthropology (bioanthropology) research. Despite several studies validating this type of research, none has examined if the assessment of dental wear magnitude based on Computerized Tomography (CT) scans is reliable. Thus, this study aims to fill this gap and assess if dental wear magnitude scoring based on CT scans provides results consistent with scoring based on direct observation of the physical specimens. MATERIALS AND METHODS: Dental wear magnitude from 412 teeth of 35 mandibles originating from the Portuguese Muge and Sado Mesolithic shell-middens was scored. The mandibles were also CT scanned and visualized using 3D Slicer. CT scan-based scoring of dental wear magnitude was then undertaken. Two scoring rounds were undertaken for each observation method (totaling four scoring rounds) and an intra-observer error test was performed. The averaged results of the two observation methods were compared via boxplots with paired cases. RESULTS: Intra-observer error was negligible and non-significant. Scoring results are comparable between the two observation methods. Notwithstanding, some differences were found, in which CT scan assessment generally overestimates dental wear when compared to direct observation. DISCUSSION: Our results generally validate the use of CT scans in studies of dental wear magnitude. Notwithstanding several caveats relating to CT scanning and visualization limitations should be considered to avoid over or under-estimation of dental wear.