Deep learning-based segmentation of malignant pleural mesothelioma tumor on computed tomography scans: application to scans demonstrating pleural effusion

Abstract

Tumor volume is a topic of interest for the prognostic assessment, treatment response evaluation, and staging of malignant pleural mesothelioma. Many mesothelioma patients present with, or develop, pleural fluid, which may complicate the segmentation of this disease. Deep convolutional neural networks (CNNs) of the two-dimensional U-Net architecture were trained for segmentation of tumor in the left and right hemithoraces, with the networks initialized through layers pretrained on ImageNet. Networks were trained on a dataset of 5230 axial sections from 154 CT scans of 126 mesothelioma patients. A test set of 94 CT sections from 34 patients, who all presented with both tumor and pleural effusion, in addition to a more general test set of 130 CT sections from 43 patients, were used to evaluate segmentation performance of the deep CNNs. The Dice similarity coefficient (DSC), average Hausdorff distance, and bias in predicted tumor area were calculated through comparisons with radiologist-provided tumor segmentations on the test sets. The present method achieved a median DSC of 0.690 on the tumor and effusion test set and achieved significantly higher performance on both test sets when compared with a previous deep learning-based segmentation method for mesothelioma.

Fully automated volumetric measurement of malignant pleural mesothelioma by deep learning AI: validation and comparison with modified RECIST response criteria

“…Nevertheless, extremely detailed ground truth was used for training, validation and comparisons between volumetry methods and readers, reducing the number of individual patients needed. The selection of 10 cases for inter-reader comparisons was arbitrary but followed precedents set in recent similar publications, including Brahim et al, 32 in which an identical number were evaluated and Sensakovic et al 33 and Gudmundsson et al 34 in which a larger number of patients, but a significantly smaller number of CT sections were interrogated. Sensakovic et al compared a total of 155 CT slices between readers (31 patients, 5 CT slices each), while Gudmundsson et al compared a total of 69 CT slices from 27 patients.…”

A Novel and Automated Approach to Classify Radiation Induced Lung Tissue Damage on CT Scans

“…The way in which a human observer deals with these challenges is by looking at adjacent slices and determining whether a pattern represents a manifestation of pathology or an acquisition artefact. That could be addressed by either incorporating 3D patches [ 38 , 39 ], which, however, comes with aforementioned challenges, or using only a few adjacent slices to predict labels for the middle slice [ 40 , 41 ], which we plan to explore in future.…”

Deep Learning for Rare Disease: A Scoping Review

“…One of the frequently used strategies for addressing the scarce data problem is data augmentation, by either modifying the original data [92 ,106-110], or generating synthetic data [62 ,64-67 ,111]. Transfer learning after pre-training of the deep learning models on other large datasets was also used to mitigate the lack of data issue [112][113][114][115][116].…”