Feasibility of moxifloxacin and proflavine dual fluorescence imaging for detecting gastrointestinal neoplastic lesions: A prospective study.

OBJECTIVES: High-contrast and high-resolution imaging techniques would enable real-time sensitive detection of the gastrointestinal lesions. This study aimed to investigate the feasibility of novel dual fluorescence imaging using moxifloxacin and proflavine in the detection of neoplastic lesions of the human gastrointestinal tract. METHODS: Patients with the colonic and gastric neoplastic lesions were prospectively enrolled. The lesions were biopsied with forceps or endoscopically resected. Dual fluorescence imaging was performed by using custom axially swept wide-field fluorescence microscopy after topical moxifloxacin and proflavine instillation. Imaging results were compared with both confocal imaging with cell labeling and conventional histological examination. RESULTS: Ten colonic samples (one normal mucosa, nine adenomas) from eight patients and six gastric samples (one normal mucosa, five adenomas) from four patients were evaluated. Dual fluorescence imaging visualized detail cellular structures. Regular glandular structures with polarized cell arrangement were observed in normal mucosa. Goblet cells were preserved in normal colonic mucosa. Irregular glandular structures with scanty cytoplasm and dispersed elongated nuclei were observed in adenomas. Goblet cells were scarce or lost in the colonic lesions. Similarity analysis between moxifloxacin and proflavine imaging showed relatively high correlation values in adenoma compared with those in normal mucosa. Dual fluorescence imaging showed good detection accuracies of 82.3% and 86.0% in the colonic and the gastric lesions, respectively. CONCLUSIONS: High-contrast and high-resolution dual fluorescence imaging was feasible for obtaining detail histopathological information in the gastrointestinal neoplastic lesions. Further studies are needed to develop dual fluorescence imaging as an in vivo real-time visual diagnostic method.

Self-Attention LSTM-FCN model for arrhythmia classification and uncertainty assessment.

This paper presents ArrhyMon, a self-attention-based LSTM-FCN model for arrhythmia classification from ECG signal inputs. ArrhyMon targets to detect and classify six different types of arrhythmia apart from normal ECG patterns. To the best of our knowledge, ArrhyMon is the first end-to-end classification model that successfully targets the classification of six detailed arrhythmia types and compared to previous work does not require additional preprocessing and/or feature extraction operations separate from the classification model. ArrhyMon's deep learning model is designed to capture and exploit both global and local features embedded in ECG sequences by integrating fully convolutional network (FCN) layers and a self-attention-based long and short-term memory (LSTM) architecture. Moreover, to enhance its practicality, ArrhyMon incorporates a deep ensemble-based uncertainty model that generates a confidence-level measure for each classification result. We evaluate ArrhyMon's effectiveness using three publicly available arrhythmia datasets (i.e., MIT-BIH, Physionet Cardiology Challenge 2017 and 2020/2021) to show that ArrhyMon achieves state-of-the-art classification performance (average accuracy 99.63%), and that confidence measures show close correlation with subjective diagnosis made from practitioners.