An interpretable and interactive deep learning algorithm for a clinically applicable retinal fundus diagnosis system by modelling finding-disease relationship.

The identification of abnormal findings manifested in retinal fundus images and diagnosis of ophthalmic diseases are essential to the management of potentially vision-threatening eye conditions. Recently, deep learning-based computer-aided diagnosis systems (CADs) have demonstrated their potential to reduce reading time and discrepancy amongst readers. However, the obscure reasoning of deep neural networks (DNNs) has been the leading cause to reluctance in its clinical use as CAD systems. Here, we present a novel architectural and algorithmic design of DNNs to comprehensively identify 15 abnormal retinal findings and diagnose 8 major ophthalmic diseases from macula-centered fundus images with the accuracy comparable to experts. We then define a notion of counterfactual attribution ratio (CAR) which luminates the system's diagnostic reasoning, representing how each abnormal finding contributed to its diagnostic prediction. By using CAR, we show that both quantitative and qualitative interpretation and interactive adjustment of the CAD result can be achieved. A comparison of the model's CAR with experts' finding-disease diagnosis correlation confirms that the proposed model identifies the relationship between findings and diseases similarly as ophthalmologists do.

Comparative analysis of high- and low-level deep learning approaches in microsatellite instability prediction.

Deep learning-based approaches in histopathology can be largely divided into two categories: a high-level approach using an end-to-end model and a low-level approach using feature extractors. Although the advantages and disadvantages of both approaches are empirically well known, there exists no scientific basis for choosing a specific approach in research, and direct comparative analysis of the two approaches has rarely been performed. Using the Cancer Genomic Atlas (TCGA)-based dataset, we compared these two different approaches in microsatellite instability (MSI) prediction and analyzed morphological image features associated with MSI. Our high-level approach was based solely on EfficientNet, while our low-level approach relied on LightGBM and multiple deep learning models trained on publicly available multiclass tissue, nuclei, and gland datasets. We compared their performance and important image features. Our high-level approach showed superior performance compared to our low-level approach. In both approaches, debris, lymphocytes, and necrotic cells were revealed as important features of MSI, which is consistent with clinical knowledge. Then, during qualitative analysis, we discovered the weaknesses of our low-level approach and demonstrated that its performance can be improved by using different image features in a complementary way. We performed our study using open-access data, and we believe this study can serve as a useful basis for discovering imaging biomarkers for clinical application.

Aggregation of cohorts for histopathological diagnosis with deep morphological analysis.

There have been substantial efforts in using deep learning (DL) to diagnose cancer from digital images of pathology slides. Existing algorithms typically operate by training deep neural networks either specialized in specific cohorts or an aggregate of all cohorts when there are only a few images available for the target cohort. A trade-off between decreasing the number of models and their cancer detection performance was evident in our experiments with The Cancer Genomic Atlas dataset, with the former approach achieving higher performance at the cost of having to acquire large datasets from the cohort of interest. Constructing annotated datasets for individual cohorts is extremely time-consuming, with the acquisition cost of such datasets growing linearly with the number of cohorts. Another issue associated with developing cohort-specific models is the difficulty of maintenance: all cohort-specific models may need to be adjusted when a new DL algorithm is to be used, where training even a single model may require a non-negligible amount of computation, or when more data is added to some cohorts. In resolving the sub-optimal behavior of a universal cancer detection model trained on an aggregate of cohorts, we investigated how cohorts can be grouped to augment a dataset without increasing the number of models linearly with the number of cohorts. This study introduces several metrics which measure the morphological similarities between cohort pairs and demonstrates how the metrics can be used to control the trade-off between performance and the number of models.

A Prospective Validation and Observer Performance Study of a Deep Learning Algorithm for Pathologic Diagnosis of Gastric Tumors in Endoscopic Biopsies.

PURPOSE: Gastric cancer remains the leading cause of cancer-related deaths in Northeast Asia. Population-based endoscopic screenings in the region have yielded successful results in early detection of gastric tumors. Endoscopic screening rates are continuously increasing, and there is a need for an automatic computerized diagnostic system to reduce the diagnostic burden. In this study, we developed an algorithm to classify gastric epithelial tumors automatically and assessed its performance in a large series of gastric biopsies and its benefits as an assistance tool. EXPERIMENTAL DESIGN: Using 2,434 whole-slide images, we developed an algorithm based on convolutional neural networks to classify a gastric biopsy image into one of three categories: negative for dysplasia (NFD), tubular adenoma, or carcinoma. The performance of the algorithm was evaluated by using 7,440 biopsy specimens collected prospectively. The impact of algorithm-assisted diagnosis was assessed by six pathologists using 150 gastric biopsy cases. RESULTS: Diagnostic performance evaluated by the AUROC curve in the prospective study was 0.9790 for two-tier classification: negative (NFD) versus positive (all cases except NFD). When limited to epithelial tumors, the sensitivity and specificity were 1.000 and 0.9749. Algorithm-assisted digital image viewer (DV) resulted in 47% reduction in review time per image compared with DV only and 58% decrease to microscopy. CONCLUSIONS: Our algorithm has demonstrated high accuracy in classifying epithelial tumors and its benefits as an assistance tool, which can serve as a potential screening aid system in diagnosing gastric biopsy specimens.

System level analysis of motor-related neural activities in larval Drosophila.

The way in which the central nervous system (CNS) governs animal movement is complex and difficult to solve solely by the analyses of muscle movement patterns. We tackle this problem by observing the activity of a large population of neurons in the CNS of larval Drosophila. We focused on two major behaviors of the larvae - forward and backward locomotion - and analyzed the neuronal activity related to these behaviors during the fictive locomotion that occurs spontaneously in the isolated CNS. We expressed a genetically-encoded calcium indicator, GCaMP and a nuclear marker in all neurons and then used digitally scanned light-sheet microscopy to record (at a fast frame rate) neural activities in the entire ventral nerve cord (VNC). We developed image processing tools that automatically detected the cell position based on the nuclear staining and allocate the activity signals to each detected cell. We also applied a machine learning-based method that we recently developed to assign motor status in each time frame. Our experimental procedures and computational pipeline enabled systematic identification of neurons that showed characteristic motor activities in larval Drosophila. We found cells whose activity was biased toward forward locomotion and others biased toward backward locomotion. In particular, we identified neurons near the boundary of the subesophageal zone (SEZ) and thoracic neuromeres, which were strongly active during an early phase of backward but not forward fictive locomotion.

Data-driven analysis of motor activity implicates 5-HT2A neurons in backward locomotion of larval Drosophila.

Rhythmic animal behaviors are regulated in part by neural circuits called the central pattern generators (CPGs). Classifying neural population activities correlated with body movements and identifying the associated component neurons are critical steps in understanding CPGs. Previous methods that classify neural dynamics obtained by dimension reduction algorithms often require manual optimization which could be laborious and preparation-specific. Here, we present a simpler and more flexible method that is based on the pre-trained convolutional neural network model VGG-16 and unsupervised learning, and successfully classifies the fictive motor patterns in Drosophila larvae under various imaging conditions. We also used voxel-wise correlation mapping to identify neurons associated with motor patterns. By applying these methods to neurons targeted by 5-HT2A-GAL4, which we generated by the CRISPR/Cas9-system, we identified two classes of interneurons, termed Seta and Leta, which are specifically active during backward but not forward fictive locomotion. Optogenetic activation of Seta and Leta neurons increased backward locomotion. Conversely, thermogenetic inhibition of 5-HT2A-GAL4 neurons or application of a 5-HT2 antagonist decreased backward locomotion induced by noxious light stimuli. This study establishes an accelerated pipeline for activity profiling and cell identification in larval Drosophila and implicates the serotonergic system in the modulation of backward locomotion.

Ultrasonically enhanced electrochemical oxidation of ibuprofen.

A hybrid advanced oxidation process combining sonochemistry (US) and electrochemistry (EC) for the batch scale degradation of ibuprofen was developed. The performance of this hybrid reactor system was evaluated by quantifying on the degradation of ibuprofen under the variation in electrolytes, frequency, applied voltage, ultrasonic power density and temperature in aqueous solutions with a platinum electrode. Among the methods examined (US, EC and US/EC), the hybrid method US/EC resulted 89.32%, 81.85% and 88.7% degradations while using NaOH, H2SO4 and deionized water (DI), respectively, with a constant electrical voltages of 30V, an ultrasound frequency of 1000kHz, and a power density of 100WL(-1) at 298K in 1h. The degradation was established to follow pseudo first order kinetics. In addition, energy consumption and energy efficiencies were also calculated. The probable mechanism for the anodic oxidation of ibuprofen at a platinum electrode was also postulated.