Autofluorescence Bronchoscopy Video Analysis for Lesion Frame Detection

Abstract

Because of the significance of bronchial lesions as indicators of early lung cancer and squamous cell carcinoma, a critical need exists for early detection of bronchial lesions. Autofluorescence bronchoscopy (AFB) is a primary modality used for bronchial lesion detection, as it shows high sensitivity to suspicious lesions. The physician, however, must interactively browse a long video stream to locate lesions, making the search exceedingly tedious and error prone. Unfortunately, limited research has explored the use of automated AFB video analysis for efficient lesion detection. We propose a robust automatic AFB analysis approach that distinguishes informative and uninformative AFB video frames in a video. In addition, for the informative frames, we determine the frames containing potential lesions and delineate candidate lesion regions. Our approach draws upon a combination of computer-based image analysis, machine learning, and deep learning. Thus, the analysis of an AFB video stream becomes more tractable. Tests with patient AFB video indicate that ≥97% of frames were correctly labeled as informative or uninformative. In addition, ≥97% of lesion frames were correctly identified, with false positive and false negative rates ≤3%.Clinical relevance-The method makes AFB-based bronchial lesion analysis more efficient, thereby helping to advance the goal of better early lung cancer detection.

A Comprehensive Study on Medical Image Segmentation using Deep Neural Networks

“…UNet is a widely used method in medical image segmentation. Its variants and hybrid network architectures have been developed to achieve higher accuracy in segmentation Unet (2015, [38]), V-net, 3D U-Net (2016, [39], [40]), Mask R-CNN (2017, [41]), U-Net++ (2018, [42]), UNet 3+, DRU-Net, DoubleU-net (2020, [43] - [45]), TransUNet, Swin Transformer (2021, [46], [47]), UNETR++, LE-UDA, TransUNet+ (2022, [48], [49], [50]), SEP (2022, [51]), ESFPNet (2022, [52])…”

“…The paper not only compares three of the most specific network architectures for MIS, but it also tracks the latest MIS rankings on paperswithcode. According to the latest updates, the two top models, SEP (2022, [51]) and ESFPNet-L (2022, [52]), are at the forefront on two datasets, Kvasir-SEG and CVC-ClinicDB, respectively. Spatially Exclusive Pasting (SEP) (2022, [51]) is an innovative data augmentation technique designed to tackle the issue of data scarcity in polyp segmentation, an important task in the diagnosis of intestinal diseases such as tumors and precancerous lesions.…”

“…The limitation of SEP is that it is only applicable to a limited number of data sources and is specifically designed for the task of polyp segmentation. ESFPNet (2022, [52]) is a deep learning architecture designed for real-time accurate segmentation and robust detection of bronchial lesions in autofluorescent bronchoscopy (AFB) video streams. Fig.…”

“…Lung cancer is usually detected early when lesions appear in the bronchial epithelium of the airway wall. Autofluorescence bronchoscopy (AFB) [52] has been shown in recent studies to be particularly effective in detecting lesions, making it a potentially crucial approach for airway evaluation. Bronchoscopy is a commonly used method for detecting early-stage lung cancer.…”

The Future of Endoscopic Navigation: A Review of Advanced Endoscopic Vision Technology

“…Recently, early clinical trials have been conducted utilizing fluoroscopy with novel algorithms in the LUNGVISION™ system [172]. Chang et al [173] utilized a CADx system with 39899 AFI video frames to detect bronchial lesions.…”