Assessing the Performance of Deep Learning for Automated Gleason Grading in Prostate Cancer.

Prostate cancer is a dominant health concern calling for advanced diagnostic tools. Utilizing digital pathology and artificial intelligence, this study explores the potential of 11 deep neural network architectures for automated Gleason grading in prostate carcinoma focusing on comparing traditional and recent architectures. A standardized image classification pipeline, based on the AUCMEDI framework, facilitated robust evaluation using an in-house dataset consisting of 34,264 annotated tissue tiles. The results indicated varying sensitivity across architectures, with ConvNeXt demonstrating the strongest performance. Notably, newer architectures achieved superior performance, even though with challenges in differentiating closely related Gleason grades. The ConvNeXt model was capable of learning a balance between complexity and generalizability. Overall, this study lays the groundwork for enhanced Gleason grading systems, potentially improving diagnostic efficiency for prostate cancer.

Serious Games in Orofacial Myofunctional Disorder Therapy for Children: An Expert Survey.

Orofacial Myofunctional Disorder (OMD) is believed to affect approximately 30-50% of all children. The various causes of OMD often revolve around an incorrect resting position of the tongue and cause symptoms such as difficulty in speech and swallowing. While these symptoms can persist and lead to jaw deformities, such as overjet and open bite, manual therapy has been shown to be effective, especially in children. However, much of the therapy must be done as home exercises by children without the supervision of a therapist. Since these exercises are often not perceived as exciting by the children, half-hearted performance or complete omission of the exercises is common, rendering the therapy less effective or completely useless. To overcome this limitation, we implemented the LudusMyo platform, a serious game platform for OMD therapy. While children are the main target group, the acceptance (and usability) assessment by experts is the first milestone for the successful implementation of an mHealth application for therapy. For this reason, we conducted an expert survey among OMD therapists to gather their input on the LudusMyo prototype. The results of this expert survey are reported in this manuscript.

Insights into the Quality of Mobile Health Apps: Preliminary Results of an Analysis of MARS Scores.

Many mHelath applications have been developed, and the Mobile App Rating Scale (MARS) is a common tool for assessing them. This study aims to provide mean values for MARS scores found in recent literature. We systematically searched for literature in which MARS was used and analyzed them. MARS values for 5,920 applications from 215 studies were compiled. The mean MARS Quality Score is 3.51. The highest average score was achieved in the Functionality category (3.98), followed by Aesthetics (3.52), Information (3.33), Engagement (3.18) and Subjective (2.72). To the best of our knowledge, this is the first study to calculate average values for the five categories of the MARS and the MARS score based on such an extensive collection of data. The study shows that the overall quality of the applications is above the average value of 2.5.

Comparing nnU-Net and deepflash2 for Histopathological Tumor Segmentation.

Machine Learning (ML) has evolved beyond being a specialized technique exclusively used by computer scientists. Besides the general ease of use, automated pipelines allow for training sophisticated ML models with minimal knowledge of computer science. In recent years, Automated ML (AutoML) frameworks have become serious competitors for specialized ML models and have even been able to outperform the latter for specific tasks. Moreover, this success is not limited to simple tasks but also complex ones, like tumor segmentation in histopathological tissue, a very time-consuming task requiring years of expertise by medical professionals. Regarding medical image segmentation, the leading AutoML frameworks are nnU-Net and deepflash2. In this work, we begin to compare those two frameworks in the area of histopathological image segmentation. This use case proves especially challenging, as tumor and healthy tissue are often not clearly distinguishable by hard borders but rather through heterogeneous transitions. A dataset of 103 whole-slide images from 56 glioblastoma patients was used for the evaluation. Training and evaluation were run on a notebook with consumer hardware, determining the suitability of the frameworks for their application in clinical scenarios rather than high-performance scenarios in research labs.

Return to work after Post-COVID: describing affected employees' perceptions of personal resources, organizational offerings and care pathways.

Background: Most individuals recover from the acute phase of infection with the SARS-CoV-2 virus, however, some encounter prolonged effects, referred to as the Post-COVID syndrome. Evidence exists that such persistent symptoms can significantly impact patients' ability to return to work. This paper gives a comprehensive overview of different care pathways and resources, both personal and external, that aim to support Post-COVID patients during their work-life reintegration process. By describing the current situation of Post-COVID patients pertaining their transition back to the workplace, this paper provides valuable insights into their needs. Methods: A quantitative research design was applied using an online questionnaire as an instrument. Participants were recruited via Post-COVID outpatients, rehab facilities, general practitioners, support groups, and other healthcare facilities. Results: The analyses of 184 data sets of Post-COVID affected produced three key findings: (1) The evaluation of different types of personal resources that may lead to a successful return to work found that particularly the individuals' ability to cope with their situation (measured with the FERUS questionnaire), produced significant differences between participants that had returned to work and those that had not been able to return so far (F = 4.913, p = 0.001). (2) In terms of organizational provisions to facilitate successful reintegration into work-life, predominantly structural changes (i.e., modification of the workplace, working hours, and task) were rated as helpful or very helpful on average (meanworkplace 2.55/SD = 0.83, meanworking hours 2.44/SD = 0.80; meantasks 2.55/SD = 0.83), while the remaining offerings (i.e., job coaching or health courses) were rated as less helpful or not helpful at all. (3) No significant correlation was found between different care pathways and a successful return to work. Conclusion: The results of the in-depth descriptive analysis allows to suggests that the level of ability to cope with the Post-COVID syndrome and its associated complaints, as well as the structural adaptation of the workplace to meet the needs and demands of patients better, might be important determinants of a successful return. While the latter might be addressed by employers directly, it might be helpful to integrate training on coping behavior early in care pathways and treatment plans for Post-COVID patients to strengthen their coping abilities aiming to support their successful return to work at an early stage.

Multiscale quantification of morphological heterogeneity with creation of a predictor of longer survival in glioblastoma.

Intratumor heterogeneity is a main cause of the dismal prognosis of glioblastoma (GBM). Yet, there remains a lack of a uniform assessment of the degree of heterogeneity. With a multiscale approach, we addressed the hypothesis that intratumor heterogeneity exists on different levels comprising traditional regional analyses, but also innovative methods including computer-assisted analysis of tumor morphology combined with epigenomic data. With this aim, 157 biopsies of 37 patients with therapy-naive IDH-wildtype GBM were analyzed regarding the intratumor variance of protein expression of glial marker GFAP, microglia marker Iba1 and proliferation marker Mib1. Hematoxylin and eosin stained slides were evaluated for tumor vascularization. For the estimation of pixel intensity and nuclear profiling, automated analysis was used. Additionally, DNA methylation profiling was conducted separately for the single biopsies. Scoring systems were established to integrate several parameters into one score for the four examined modalities of heterogeneity (regional, cellular, pixel-level and epigenomic). As a result, we could show that heterogeneity was detected in all four modalities. Furthermore, for the regional, cellular and epigenomic level, we confirmed the results of earlier studies stating that a higher degree of heterogeneity is associated with poorer overall survival. To integrate all modalities into one score, we designed a predictor of longer survival, which showed a highly significant separation regarding the OS. In conclusion, multiscale intratumor heterogeneity exists in glioblastoma and its degree has an impact on overall survival. In future studies, the implementation of a broadly feasible heterogeneity index should be considered.

Brain Tumor Classification and Segmentation Using Dual-Outputs for U-Net Architecture: O2U-Net.

We propose a modified version of the U-Net architecture for segmenting and classifying brain tumors, introducing another output between down- and up-sampling. Our proposed architecture utilizes two outputs, adding a classification output beside the segmentation output. The central idea is to use fully connected layers to classify each image before applying U-Net's up-sampling operations. This is achieved by utilizing the features extracted during the down-sampling procedure and combining them with fully connected layers for classification. Afterward, the segmented image is generated by U-Net's up-sampling process. Initial tests show competitive results against comparable models with 80.83%, 99.34%, and 77.39% for the dice coefficient, accuracy, and sensitivity, respectively. The tests were conducted on the well-established dataset from Nanfang Hospital, Guangzhou, China, and General Hospital, Tianjin Medical University, China, from 2005 to 2010 containing MRI images of 3064 brain tumors.

Evaluation of Completeness, Comparability, Validity, and Timeliness in Cancer Registries: A Scoping Review.

An essential aspect of cancer registration is data quality. Data quality for Cancer Registries has been reviewed in this paper using four main criteria (comparability, validity, timeliness, and completeness). Medline (via PubMed), Scopus, and Web of Science databases were searched for relevant English articles published from inception until December 2022. Each study was analyzed for its characteristics, measurement method, and data quality features. According to the present study, the majority of articles evaluated the completeness feature, and the fewest evaluated the timeliness feature. A completeness rate of 36% to 99.3% and a timeliness rate of 9% to 98.5% were observed. Standardizing metrics and reporting of data quality is necessary to maintain confidence in the usefulness of cancer registries.

Machine Learning Approaches for Detecting Coronary Artery Disease Using Angiography Imaging: A Scoping Review.

This scoping review aims to identify and summarize the current literature on Machine learning (ML) approaches for detecting coronary artery disease (CAD) using angiography imaging. We comprehensively searched several databases and identified 23 studies that met the inclusion criteria. They employed different types of angiography imaging including computed tomography and invasive coronary angiography. Several studies have used deep learning algorithms for image classification and segmentation, and our findings show that various machine learning algorithms, such as convolutional neural networks, different types of U-Net, and hybrid approaches. Studies also varied in the outcomes measured, identifying stenosis, and assessing the severity of CAD. ML approaches can improve the accuracy and efficiency of CAD detection by using angiography. The performance of the algorithms differed depending on the dataset used, algorithm employed, and features selected for analysis. Therefore, there is a need to develop ML tools that can be easily integrated into clinical practice to aid in the diagnosis and management of CAD.