Value of artificial intelligence with novel tumor tracking technology in the diagnosis of gastric submucosal tumors by contrast-enhanced harmonic endoscopic ultrasonography.

BACKGROUND AND AIM: Contrast-enhanced harmonic endoscopic ultrasonography (CH-EUS) is useful for the diagnosis of lesions inside and outside the digestive tract. This study evaluated the value of artificial intelligence (AI) in the diagnosis of gastric submucosal tumors by CH-EUS. METHODS: This retrospective study included 53 patients with gastrointestinal stromal tumors (GISTs) and leiomyomas, all of whom underwent CH-EUS between June 2015 and February 2020. A novel technology, SiamMask, was used to track and trim the lesions in CH-EUS videos. CH-EUS was evaluated by AI using deep learning involving a residual neural network and leave-one-out cross-validation. The diagnostic accuracy of AI in discriminating between GISTs and leiomyomas was assessed and compared with that of blind reading by two expert endosonographers. RESULTS: Of the 53 patients, 42 had GISTs and 11 had leiomyomas. Mean tumor size was 26.4 mm. The consistency rate of the segment range of the tumor image extracted by SiamMask and marked by the endosonographer was 96% with a Dice coefficient. The sensitivity, specificity, and accuracy of AI in diagnosing GIST were 90.5%, 90.9%, and 90.6%, respectively, whereas those of blind reading were 90.5%, 81.8%, and 88.7%, respectively (P = 0.683). The κ coefficient between the two reviewers was 0.713. CONCLUSIONS: The diagnostic ability of CH-EUS results evaluated by AI to distinguish between GISTs and leiomyomas was comparable with that of blind reading by expert endosonographers.

Clinical Application of the FoundationOne CDx Assay to Therapeutic Decision-Making for Patients with Advanced Solid Tumors.

BACKGROUND: Implementation of personalized medicine requires the accessibility of tumor molecular profiling in order to allow prioritization of appropriate targeted therapies for individual patients. Our aim was to study the role of comprehensive genomic profiling assays that may inform treatment recommendations for patients with solid tumors. MATERIALS AND METHODS: We performed a prospective study to evaluate the feasibility of application of the FoundationOne CDx panel-which detects substitutions, insertions and deletions, and copy number alterations in 324 genes, select gene rearrangements, and genomic signatures including microsatellite instability and tumor mutation burden (TMB)-to patients with advanced or recurrent solid tumors before its approval in Japan. RESULTS: A total of 181 samples were processed for genomic testing between September 2018 and June 2019, with data being successfully obtained for 175 of these samples, yielding a success rate of 96.7%. The median turnaround time was 41 days (range, 21-126 days). The most common known or likely pathogenic variants were TP53 mutations (n = 113), PIK3CA mutations (n = 33), APC mutations (n = 32), and KRAS mutations (n = 29). Among the 153 patients assessed for TMB, the median TMB was 4 mutations/Mb, and tumors with a high TMB (≥10 mutations/Mb) were more prevalent for lung cancer (11/32) than for other solid tumor types (9/121, Fisher's exact test p < .01). No clear trend toward increased efficacy for immune checkpoint inhibitor (ICI) monotherapy or ICI combination chemotherapy in patients with a high programmed cell death-ligand 1 tumor proportion score or a high TMB was apparent. Among the 174 patients found to harbor known or likely pathogenic actionable alterations, 24 individuals (14%) received matched targeted therapy. CONCLUSION: The FoundationOne CDx assay was performed with formalin-fixed, paraffin-embedded tumor specimens with a success rate of >95%. Such testing may inform the matching of patients with cancer with investigational or approved targeted drugs. IMPLICATIONS FOR PRACTICE: This prospective cohort study was initiated to investigate the feasibility and utility of clinical application of FoundationOne CDx. A total of 181 samples were processed for genomic testing between September 2018 and June 2019, with data being successfully obtained for 175 of these samples, yielding a success rate of 96.7%, and 24 individuals (14%) received matched targeted therapy.

Computer-Aided Diagnosis Based on Convolutional Neural Network System for Colorectal Polyp Classification: Preliminary Experience.

BACKGROUND AND AIM: Computer-aided diagnosis (CAD) is becoming a next-generation tool for the diagnosis of human disease. CAD for colon polyps has been suggested as a particularly useful tool for trainee colonoscopists, as the use of a CAD system avoids the complications associated with endoscopic resections. In addition to conventional CAD, a convolutional neural network (CNN) system utilizing artificial intelligence (AI) has been developing rapidly over the past 5 years. We attempted to generate a unique CNN-CAD system with an AI function that studied endoscopic images extracted from movies obtained with colonoscopes used in routine examinations. Here, we report our preliminary results of this novel CNN-CAD system for the diagnosis of colon polyps. METHODS: A total of 1,200 images from cases of colonoscopy performed between January 2010 and December 2016 at Kindai University Hospital were used. These images were extracted from the video of actual endoscopic examinations. Additional video images from 10 cases of unlearned processes were retrospectively assessed in a pilot study. They were simply diagnosed as either an adenomatous or nonadenomatous polyp. RESULTS: The number of images used by AI to learn to distinguish adenomatous from nonadenomatous was 1,200:600. These images were extracted from the videos of actual endoscopic examinations. The size of each image was adjusted to 256 × 256 pixels. A 10-hold cross-validation was carried out. The accuracy of the 10-hold cross-validation is 0.751, where the accuracy is the ratio of the number of correct answers over the number of all the answers produced by the CNN. The decisions by the CNN were correct in 7 of 10 cases. CONCLUSION: A CNN-CAD system using routine colonoscopy might be useful for the rapid diagnosis of colorectal polyp classification. Further prospective studies in an in vivo setting are required to confirm the effectiveness of a CNN-CAD system in routine colonoscopy.

Utility of U-Net for the objective segmentation of the fibroglandular tissue region on clinical digital mammograms.

This study investigates the equivalence or compatibility between U-Net and visual segmentations of fibroglandular tissue regions by mammography experts for calculating the breast density and mean glandular dose (MGD). A total of 703 mediolateral oblique-view mammograms were used for segmentation. Two region types were set as the ground truth (determined visually): (1) one type included only the region where fibroglandular tissue was identifiable (called the 'dense region'); (2) the other type included the region where the fibroglandular tissue may have existed in the past, provided that apparent adipose-only parts, such as the retromammary space, are excluded (the 'diffuse region'). U-Net was trained to segment the fibroglandular tissue region with an adaptive moment estimation optimiser, five-fold cross-validated with 400 training and 100 validation mammograms, and tested with 203 mammograms. The breast density and MGD were calculated using the van Engeland and Dance formulas, respectively, and compared between U-Net and the ground truth with the Dice similarity coefficient and Bland-Altman analysis. Dice similarity coefficients between U-Net and the ground truth were 0.895 and 0.939 for the dense and diffuse regions, respectively. In the Bland-Altman analysis, no proportional or fixed errors were discovered in either the dense or diffuse region for breast density, whereas a slight proportional error was discovered in both regions for the MGD (the slopes of the regression lines were -0.0299 and -0.0443 for the dense and diffuse regions, respectively). Consequently, the U-Net and ground truth were deemed equivalent (interchangeable) for breast density and compatible (interchangeable following four simple arithmetic operations) for MGD. U-Net-based segmentation of the fibroglandular tissue region was satisfactory for both regions, providing reliable segmentation for breast density and MGD calculations. U-Net will be useful in developing a reliable individualised screening-mammography programme, instead of relying on the visual judgement of mammography experts.

Artificial intelligence-based endoscopic diagnosis of colorectal polyps using residual networks.

Convolutional neural networks (CNNs) are widely used for artificial intelligence (AI)-based image classification. Residual network (ResNet) is a new technology that facilitates the accuracy of image classification by CNN-based AI. In this study, we developed a novel AI model combined with ResNet to diagnose colorectal polyps. In total, 127,610 images consisting of 62,510 images with adenomatous polyps, 30,443 with non-adenomatous hyperplastic polyps, and 34,657 with healthy colorectal normal mucosa were subjected to deep learning after annotation. Each validation process was performed using 12,761 stored images of colorectal polyps by a 10-fold cross validation. The efficacy of the ResNet system was evaluated by sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy. The sensitivity, specificity, PPV, NPV, and diagnostic accuracy for adenomatous polyps at WLIs were 98.8%, 94.3%, 90.5%, 87.4%, and 92.8%, respectively. Similar results were obtained for adenomatous polyps at narrow-band imagings (NBIs) and chromoendoscopy images (CEIs) (NBIs vs. CEIs: sensitivity, 94.9% vs. 98.2%; specificity, 93.9% vs. 85.8%; PPV, 92.5% vs. 81.7%; NPV, 93.5% vs. 99.9%; and overall accuracy, 91.5% vs. 90.1%). The ResNet model is a powerful tool that can be used for AI-based accurate diagnosis of colorectal polyps.

The effectiveness of mutation operation in the case of Estimation of Distribution Algorithms.

The Estimation of Distribution Algorithms are a class of evolutionary algorithms which adopt probabilistic models to reproduce individuals in the next generation, instead of conventional crossover and mutation operators. In this paper, mutation operators are incorporated into Estimation of Distribution Algorithms in order to maintain the diversities in EDA populations. Two kinds of mutation operators are examined: a bitwise mutation operator and a mutation operator taking account into the probabilistic model. In experiments, we do not only compare the proposed methods with conventional EDAs on a few fitness functions but also analyze sampled probabilistic models by using KL-divergence. The experimental results shown in this paper elucidate that the mutation operator taking account into the probabilistic model improve the search ability of EDAs.