Search

Background: Bladder cancer (BCa) is the second most common type of cancer in the genitourinary system and causes approximately 165,000 deaths each year. The diagnosis of BCa is primarily done through cystoscopy, which involves visually examining the bladder using an endoscope. Currently, white light cystoscopy is considered the most reliable method for diagnosis. However, it can be challenging to detect and diagnose flat, small, or poorly textured lesions. The study explores the performance of deep learning systems (YOLOv7- tiny, RT-DETR18), originally designed for detecting adenomas in colonoscopy images, when retrained and tested with cystoscopy images. The deep neural network used in the study was pre-trained on 35,699 colonoscopy images (some from Mannheim) and both architectures achieved a F1 score of 0.91 on publicly available colonoscopy datasets. Results: When the adenoma-detection network was tested with cystoscopy images from two sources (Ulm and Erlangen), F1 scores ranging from 0.58 to 0.81 were achieved. Subsequently, the networks were retrained using 12,066 cystoscopy images from Mannheim, resulting in improved F1 scores ranging from 0.77 to 0.85. Conclusion: It could be shown that transformer based networks perform slightly better than YOLOv7-tiny networks, but both network types are feasable for lesion detection in the human bladder. The retraining of the network with additional cystoscopy data led to an improvement in the performance of urinary lesion detection. This suggests that it is possible to achieve a domain-shift with the inclusion of appropriate additional data.

Background: Bladder cancer (BCa) is the second most common genitourinary malignancy and has a mortality of 165,000 deaths p.a. The diagnosis of BCa is mostly carried out using cystoscopy - the visual examination of the urinary bladder with an endoscope. White light cystoscopy is currently considered as gold standard for the diagnosis. Nevertheless, especially flat, small or weakly textured lesions, are very difficult to detect and diagnose. Objective: With the advent of deep learning and already commercially available systems for the detection of adenomas in colonoscopy, it is investigated how such a system - for colonoscopy - performs if retrained and tested with cystoscopy images. Methods: A deep neural network with a YOLOv7-tiny architecture was pre-trained on 35,699 colonoscopy images (partially from Mannheim), yielding a precision = 0.92, sensitivity = 0.90, F1 = 0.91 on public colonoscopy data collections. Results: Testing this adenomadetection network with cystoscopy images from three sources (Ulm, Erlangen, Pforzheim), F1 scores in the range of 0.67 to 0.74 could be achieved. The network was then retrained with 12,066 cystoscopy images (from Mannheim), yielding improved F1 scores in the range of 0.78 to 0.85. Conclusion: It could be shown that a deep learning network for adenoma detection in colonoscopy is ad-hoc able to detect approximately 75% of the lesions in the urinary bladder in cystoscopy images, suggesting that these lesions have a similar appearance. After retraining the network with additional cystoscopy data, the performance for urinary lesion detection could be improved, indicating that a domain-shift with adequate additional data is feasible.

Introduction: Early detection of adenomas and polyps is one central goal of colonoscopic screening programs. As the adenoma detection rate (ADR) depends on the experience of the endoscopist, AI-based polyp detection systems can be used for real-time assistance. Hence, to support the physicians such AI-based systems using deep-convolutional neural networks (DCNNS) have been introduced in the past years. One disadvantage of these techniques is the need of a huge amount of labeled training data. Method: We investigate a "human-in-the-loop approach" to minimize the required time to generate labeled training data. The approach is evaluated within the training a YOLOv4 neural network to detect polyps in colonoscopic image data. The performance metrics of the neural network are evaluated on three public datasets. Results: The performance of the YOLO network increased from a precision of 0.88, recall of 0.83, F1 score of 0.86, and a F2 score of 0.86 to a precision of 0.91, recall of 0.87, F1 = of 0.89, and F2 = of 0.88. The interactive labelling of 1,000 images only takes one hours. Conclusion: The proposed "human-in-theloop approach" is capable of generating labelled image data in a minimum of time while increasing the performance metrics as well. For higher performance increase more data can now be labeled within this new approach.

Introduction: Postprocedural bleeding is a major adverse event after endoscopic resection of colorectal lesions, but the optimal surveillance time after endoscopy is unclear. In this study, we determined onset time and characteristics of postprocedural bleeding events. Methods: We retrospectively screened patients who underwent endoscopic resection of colorectal lesions at three German hospitals between 2010 and 2019 for postprocedural bleeding events using billing codes. Only patients who required re-endoscopy were included for analysis. For identified patients, we collected demographic data, clinical courses, characteristics of colorectal lesions, and procedure-related variables. Factors associated with late-onset bleeding were determined by univariate and multivariate logistic regression analysis. Results: From a total of 6,820 patients with eligible billing codes, we identified 113 cases with postprocedural bleeding after endoscopic mucosal (61.9%) or snare resection (38.1%) that required re-endoscopy. The median size of the culprit lesion was 20 mm (interquartile range 14–30 mm). The median onset time of postprocedural bleeding was day 3 (interquartile range: 1–6.5 days), with 48.7% of events occurring within 48 h. Multivariate logistic regression analysis demonstrates that a continued intake of antiplatelet drugs (OR: 3.98, 95% CI: 0.89–10.12, p = 0.025) and a flat morphology of the colorectal lesion (OR: 2.98, 95% CI: 1.08–8.01, p = 0.031) were associated with an increased risk for late postprocedural bleeding (>48 h), whereas intraprocedural bleeding was associated with a decreased risk (OR: 0.12, 95% CI: 0.04–0.50, p = 0.001). Conclusion: Significant postprocedural bleeding can occur up to 18 days after endoscopic resection of colorectal lesions, but was predominantly observed within 48 h. Continued intake of antiplatelet drugs and a flat polyp morphology are associated with risk for late postprocedural bleeding. [ABSTRACT FROM AUTHOR]

Background: Bladder cancer (BCa) is the second most common genitourinary malignancy and has a mortality of 165,000 deaths p.a. The diagnosis of BCa is mostly carried out using cystoscopy - the visual examination of the urinary bladder with an endoscope. White light cystoscopy is currently considered as gold standard for the diagnosis. Nevertheless, especially flat, small or weakly textured lesions, are very difficult to detect and diagnose. Objective: With the advent of deep learning and already commercially available systems for the detection of adenomas in colonoscopy, it is investigated how such a system - for colonoscopy - performs if retrained and tested with cystoscopy images. Methods: A deep neural network with a YOLOv7-tiny architecture was pre-trained on 35,699 colonoscopy images (partially from Mannheim), yielding a precision = 0.92, sensitivity = 0.90, F1 = 0.91 on public colonoscopy data collections. Results: Testing this adenomadetection network with cystoscopy images from three sources (Ulm, Erlangen, Pforzheim), F1 scores in the range of 0.67 to 0.74 could be achieved. The network was then retrained with 12,066 cystoscopy images (from Mannheim), yielding improved F1 scores in the range of 0.78 to 0.85. Conclusion: It could be shown that a deep learning network for adenoma detection in colonoscopy is ad-hoc able to detect approximately 75% of the lesions in the urinary bladder in cystoscopy images, suggesting that these lesions have a similar appearance. After retraining the network with additional cystoscopy data, the performance for urinary lesion detection could be improved, indicating that a domain-shift with adequate additional data is feasible. [ABSTRACT FROM AUTHOR]

Benchmarking programs for crop and livestock farms have been used by farmers to identify ways to improve farm efficiencies. This study developed a set of benchmarks for oyster, mussel, and seaweed farming in Maine (USA). Farm-level survey data were used to calculate benchmarking metrics for each farm respondent. Results showed the importance of disaggregating benchmarking metrics by production scale, gear type, and whether farms were initial startup (defined in this study as <5 years in business) or established (>5 years in business) farms. The initial analysis of cost structures on individual farms provided useful information from which to identify key categories for disaggregation. Identification of the most important cost inputs further points to which benchmarks will be of greatest value for different groups of farms. Results for oyster farms showed that startup costs per oyster harvested were lowest ($0.20/oyster) on established bottom culture farms as compared to $0.96/oyster for established suspended oyster farms, and highest ($1.75/oyster) on startup oyster farms. Profits (Net Farm Income) per oyster were greatest on established bottom oyster farms followed by established suspended oyster farms, but were negative on average, for startup farms. Per-hectare, however, profits were greater on established suspended than bottom culture oyster farms. Labor efficiencies were also greatest for established suspended oyster farms (65 oysters/hour of labor), followed by established bottom culture oyster farms (55 oysters/hour of labor), with the lowest labor efficiency (44 oysters/hour of labor) for startup oyster farms. Given the inherent variability among aquaculture farms, adequate numbers of observations of participating farmers are necessary within each production scale/gear type and startup/established categories for benchmarks to be of value. For emerging sectors of aquaculture, benchmarking metrics can be useful for navigating the critical startup period, but obtaining sufficient numbers of observations is a challenge. Benchmarking metrics for established farm businesses provide guidance on the levels of production and economic performance necessary to be successful. Overall, benchmark values are most useful when applied in a holistic fashion that takes into consideration the performance of the farm across all production, revenue, expense, and efficiency categories (labor, capital, and financial). [ABSTRACT FROM AUTHOR]

Sustainable domestic aquaculture development is a critical component to achieving greater U.S. seafood security in the future, yet detrimental allegations have corrupted public support. A variety of longstanding and inaccurate myths and assumptions directed at offshore aquaculture farming and its regulation have been foisted on the public. This paper refutes the most prevalent critiques by reviewing current policies, regulations, research and industry production practices. These criticisms include: inadequate regulatory oversight; portrayal of farms as being high density factories unconcerned by food waste, untreated discharge, use of antibiotic and antifungal treatments; entanglement of marine mammals; impacts on wild stocks and habitats; use of feed additives to pigment fish flesh; unsustainable use of fish meal in feed formulations; potential market disruption by producing cheap, low quality products; and commercial farms and commercial fishers cannot coexist as for-profit businesses. Marine aquaculture is not risk-free in terms of potential environmental, economic, social, and cultural impacts and challenges remain to achieve a sustainable industry. These challenges are well known and addressable by the U.S. and global research community. Current offshore farming realities bode well for the future: 1) there is a clear global imperative to sustainably produce more seafood to meet growing demand and the U.S. has the marine resources to become a major exporter, if U.S. law can be amended to grant offshore farmers a property right or security of tenure for sites in federal waters; 2) U.S. ocean farmers work within a very complex and effective legal, regulatory, science-driven environment to anticipate and mitigate potential impacts; 3) farm level management decisions and federal and state regulatory frameworks have worked together to bring about environmentally friendly siting, operational, and production outcomes, and 4) the farming community and its advocates in government, universities, and industry recognize it is essential to reach out to decision-makers and the interested public, as well as critics, with the latest research and empirical results to present an accurate picture of risks and rewards to development. [ABSTRACT FROM AUTHOR]