Effect of Training Data Volume on Performance of Convolutional Neural Network Pneumothorax Classifiers.

Large datasets with high-quality labels required to train deep neural networks are challenging to obtain in the radiology domain. This work investigates the effect of training dataset size on the performance of deep learning classifiers, focusing on chest radiograph pneumothorax detection as a proxy visual task in the radiology domain. Two open-source datasets (ChestX-ray14 and CheXpert) comprising 291,454 images were merged and convolutional neural networks trained with stepwise increase in training dataset sizes. Model iterations at each dataset volume were evaluated on an external test set of 525 emergency department chest radiographs. Learning curve analysis was performed to fit the observed AUCs for all models generated. For all three network architectures tested, model AUCs and accuracy increased rapidly from 2 × 103 to 20 × 103 training samples, with more gradual increase until the maximum training dataset size of 291 × 103 images. AUCs for models trained with the maximum tested dataset size of 291 × 103 images were significantly higher than models trained with 20 × 103 images: ResNet-50: AUC20k = 0.86, AUC291k = 0.95, p < 0.001; DenseNet-121 AUC20k = 0.85, AUC291k = 0.93, p < 0.001; EfficientNet AUC20k = 0.92, AUC 291 k = 0.98, p < 0.001. Our study established learning curves describing the relationship between dataset training size and model performance of deep learning convolutional neural networks applied to a typical radiology binary classification task. These curves suggest a point of diminishing performance returns for increasing training data volumes, which algorithm developers should consider given the high costs of obtaining and labelling radiology data.

Restaging of rectal cancer with hybrid positron emission tomography magnetic resonance imaging after preoperative chemoradiotherapy.

Introduction: This study determines the sensitivity and specificity of positron emission tomography/magnetic resonance imaging (PET/MRI) parameters in predicting treatment response in patients with localised rectal cancer who have undergone preoperative chemoradiotherapy (CRT). Method: Patients with stage I-III adenocarcinoma of the rectum planned for preoperative CRT followed by surgery were recruited. Patients had PET/MRI scans at baseline and 6-8 weeks post-CRT. Functional MRI and PET parameters were assessed for their diagnostic accuracy for tumour regression grade (TRG). Nonparametric receiver operating characteristic analysis was employed to determine the area under the ROC curve (AUC), and the sensitivity and specificity of each quantile cut-off. Results: A total of 31 patients were recruited, of whom 20 completed study protocol. All patients included had mid or lower rectal tumours. There were 16 patients (80%) with node-positive disease at presentation. The median time to surgery was 75.5 days (range 52-106 days). Histopathological assessment revealed 20% good responders (TRG 1/2), and the remaining 80% of patients had a poor response (TRG 3/4). When predicting good responders, the AUC values for percent maximum thickness reduction and percent apparent diffusion coefficient (ADC) change were 0.82 and 0.73, respectively. A maximum thickness reduction cut-off of >47% and a percent ADC change of >20% yielded a sensitivity and specificity of 75%/95% and 75%/73%, respectively. Conclusion: Parameters such as percent maximum thickness reduction and percent ADC change may be useful for predicting good responders in patients undergoing preoperative CRT for rectal cancer. Larger studies are warranted to establish the utility of PET/MRI in rectal cancer staging.