Diagnostic Software Research Articles

Raman Spectroscopy for Instant Bladder Tumor Diagnosis: System Development and In Vivo Proof-Of-Principle Study in Accordance with the European Medical Device Regulation (MDR2017/745).

This work reports on an in vivo Raman-based endoscopy system, invaScope, enabling Raman measurements of healthy and tumor bladder tissue during an endoscopic procedure in the operating theatre. The presented study outlines the progression from the initial concept (validated through previously performed ex vivo studies) to the approval and implementation of a clinical investigational device according to the requirement within the framework of the European Medical Device Regulation (MDR2017/745). The study's primary objective was to employ the invaScope Raman system within the bladder, capturing in vivo spectroscopic Raman data followed by standard histo- and cytopathological examinations of urological tissue (considered the gold standard). The collected data were analyzed and correlated with histopathological findings post-procedure. Additionally, the study aimed to assess the feasibility of using diagnostic equipment, probes, and software for application in a clinical setting, evaluating usability aspects that are important during surgical procedures. This research represents a pivotal step toward advancing Raman spectroscopy for routine clinical use in characterizing bladder lesions.

Advancements and Best Practices in Hard Disk Drives (HDDs) Repair: A Comprehensive Review of Diagnostic Techniques, Repair Methods, and Preventive Strategies

Hard Disk Drives (HDDs) remain a crucial component of data storage systems despite the advent of Solid State Drives (SSDs). However, the mechanical and electronic nature of HDDs makes them susceptible to various failures over time. This article provides a comprehensive examination of HDD repair, focusing on the anatomy and function of these devices, common causes of failure, diagnostic techniques, and effective repair methods. Mechanical failures, such as head crashes and spindle motor issues, electronic failures like PCB damage, and logical failures including file system corruption and bad sectors are explored in detail. Diagnostic approaches utilizing visual inspection, diagnostic software tools, hardware-based diagnostics, and SMART monitoring are discussed to aid in identifying issues accurately. Repair methods covered include data recovery techniques, mechanical repairs, electronic component replacement, and firmware fixes. Real-world case studies illustrate practical challenges and solutions in HDD repair, highlighting the complexities involved. Additionally, best practices and safety measures for handling and maintaining HDDs are provided to prevent future failures and ensure data integrity. This article aims to equip data recovery professionals, IT technicians, and tech enthusiasts with the knowledge and skills necessary to diagnose and repair HDD issues effectively. By integrating technical insights and practical approaches, it serves as a valuable resource for enhancing the reliability and longevity of HDDs in various applications.

Barriers and facilitators to implementing imaging-based diagnostic artificial intelligence-assisted decision-making software in hospitals in China: a qualitative study using the updated Consolidated Framework for Implementation Research

ObjectivesTo identify the barriers and facilitators to the successful implementation of imaging-based diagnostic artificial intelligence (AI)-assisted decision-making software in China, using the updated Consolidated Framework for Implementation Research (CFIR) as...

Влияние нарушений качества рентгенографических исследований органов грудной клетки на работу врачей-рентгенологов и диагностического искусственного интеллекта.

When performing radiographic studies, quality violations and errors periodically occur, making it difficult for an image to be interpreted and described by both a radiologist and software based on artificial intelligence technology. Incorrect filling of meta-information in DICOM format headers may be a separate problem for diagnostic artificial intelligence. Purpose: determination of quality violations of the chest x-ray examinations, which most strongly impede the work of radiologists and diagnostic software based on artificial intelligence in the conditions of the Moscow city health care system. Material and methods. To study the impact of the quality of x-ray examinations on the work of a radiologist, an online survey was conducted among radiologists, employees of the Moscow reference center for radiation diagnostics. To study the impact of the quality of x-ray examinations on the work of AI-based software, an analysis of the results of processing diagnostic studies according to the modalities chest x-ray and lung fluorography for the 4th quarter of 2023 by artificial intelligence services was carried out as part of an “Experiment on the use of innovative computer vision technologies for medical image analysis and subsequent applicability in the healthcare system of Moscow”. Results. The survey involved 172 radiologists. The most common quality violations by doctors are incompleteness of chest coverage and asymmetric position of the patient, as well as violations of image clarity / contrast. At the same time, the incompleteness of the chest coverage causes the greatest difficulties in interpreting the study for doctors with up to 5 years of experience. For more experienced doctors, the greatest difficulties are caused by violations of the clarity and contrast of images. The main problems hindering the description of studies by AI-services are non-filling or incorrect filling of meta-information about the study stored in DICOM format, as well as violations of the patient’s laying and positioning. Conclusion. The problems we have identified indicate the need for more careful adherence to the methodology of conducting diagnostic studies, especially with regard to filling out meta-information about the study. AI-based software developers need to evaluate the performance of their solutions based on the developed testing and monitoring methodology.

Artificial intelligence software for assessing brain ischemic penumbra/core infarction on computed tomography perfusion: A real-world accuracy study.

With the increasingly extensive application of artificial intelligence (AI) in medical systems, the accuracy of AI in medical diagnosis in the real world deserves attention and objective evaluation. To investigate the accuracy of AI diagnostic software (Shukun) in assessing ischemic penumbra/core infarction in acute ischemic stroke patients due to large vessel occlusion. From November 2021 to March 2022, consecutive acute stroke patients with large vessel occlusion who underwent mechanical thrombectomy (MT) post-Shukun AI penumbra assessment were included. Computed tomography angiography (CTA) and perfusion exams were analyzed by AI, reviewed by senior neurointerventional experts. In the case of divergences among the three experts, discussions were held to reach a final conclusion. When the results of AI were inconsistent with the neurointerventional experts' diagnosis, the diagnosis by AI was considered inaccurate. A total of 22 patients were included in the study. The vascular recanalization rate was 90.9%, and 63.6% of patients had modified Rankin scale scores of 0-2 at the 3-month follow-up. The computed tomography (CT) perfusion diagnosis by Shukun (AI) was confirmed to be invalid in 3 patients (inaccuracy rate: 13.6%). AI (Shukun) has limits in assessing ischemic penumbra. Integrating clinical and imaging data (CT, CTA, and even magnetic resonance imaging) is crucial for MT decision-making.

AFFnet - a deep convolutional neural network for the detection of atypical femur fractures from anteriorposterior radiographs

Despite well-defined criteria for radiographic diagnosis of atypical femur fractures (AFFs), missed and delayed diagnosis is common. An AFF diagnostic software could provide timely AFF detection to prevent progression of incomplete or development of contralateral AFFs. In this study, we investigated the ability for an artificial intelligence (AI)-based application, using deep learning models (DLMs), particularly convolutional neural networks (CNNs), to detect AFFs from femoral radiographs. A labelled Australian dataset of pre-operative complete AFF (cAFF), incomplete AFF (iAFF), typical femoral shaft fracture (TFF), and non-fractured femoral (NFF) X-ray images in anterior-posterior view were used for training (N = 213, 49, 394, 1359, respectively). An AFFnet model was developed using a pretrained (ImageNet dataset) ResNet-50 backbone, and a novel Box Attention Guide (BAG) module to guide the model's scanning patterns to enhance its learning. All images were used to train and internally test the model using a 5-fold cross validation approach, and further validated by an external dataset. External validation of the model's performance was conducted on a Sweden dataset comprising 733 TFF and 290 AFF images. Precision, sensitivity, specificity, F1-score and AUC were measured and compared between AFFnet and a global approach with ResNet-50. Excellent diagnostic performance was recorded in both models (all AUC >0.97), however AFFnet recorded lower number of prediction errors, and improved sensitivity, F1-score and precision compared to ResNet-50 in both internal and external testing. Sensitivity in the detection of iAFF was higher for AFFnet than ResNet-50 (82 % vs 56 %). In conclusion, AFFnet achieved excellent diagnostic performance on internal and external validation, which was superior to a pre-existing model. Accurate AI-based AFF diagnostic software has the potential to improve AFF diagnosis, reduce radiologist error, and allow urgent intervention, thus improving patient outcomes.

Current situation and future issues with Kampo medicine: A survey of Japanese physicians

AbstractAimWe aimed to elucidate the present clinical situation, future challenges, and potential need for diagnostic support software in Kampo medicine.MethodsWe conducted an online survey (comprising 28 questions) of Japanese physicians from the 23 to 24 January 2024 via MedPeer. The participants' responses were stratified according to their age distribution.ResultsWe evaluated data for 652 respondents. The most common respondent age groups were those in their 30s and 40s. The most common specialty of respondents was internal medicine (43.7%). Only 33.7% of respondents had received education on Kampo medicine at university. Most respondents (90.0%) were not licensed as certified physicians or members of the Japan Society for Oriental Medicine. Most respondents (86.7%) currently prescribed Kampo formulas (current users); only 3.8% had no experience in prescribing Kampo formulas. The conditions for which Kampo formulas were frequently prescribed were muscle cramps, constipation, medically unexplained symptoms/menopausal syndrome, appetite loss/malnutrition, and fatigue. However, 48.3% of current users did not use traditional Kampo patterns (sho). The most frequent issues reported by respondents were the lack of scientific evidence and standardization of diagnostic methods in Kampo medicine. Respondents indicated the need for postgraduate education and implementation of diagnostic support software for using Kampo patterns by non‐specialists in Kampo medicine.ConclusionThere was a discrepancy between the high frequency of Kampo formula prescriptions and the low use of traditional Kampo patterns in Japan. Diagnostic support software is likely to become a useful method of improving appropriate Kampo formula prescriptions by non‐specialists.

A deep learning approach to detection of oral cancer lesions from intra oral patient images: A preliminary retrospective study

IntroductionOral squamous cell carcinomas (OSCC) seen in the oral cavity are a category of diseases for which dentists may diagnose and even cure. This study evaluated the performance of diagnostic computer software developed to detect oral cancer lesions in intra-oral retrospective patient images. Materials and methodsOral cancer lesions were labeled with CranioCatch labeling program (CranioCatch, Eskişehir, Turkey) and polygonal type labeling method on a total of 65 anonymous retrospective intraoral patient images of oral mucosa that were diagnosed with oral cancer histopathologically by incisional biopsy from individuals in our clinic. All images have been rechecked and verified by experienced experts. This data set was divided into training (n = 53), validation (n = 6) and test (n = 6) sets. Artificial intelligence model was developed using YOLOv5 architecture, which is a deep learning approach. Model success was evaluated with confusion matrix. ResultsWhen the success rate in estimating the images reserved for the test not used in education was evaluated, the F1, sensitivity and precision results of the artificial intelligence model obtained using the YOLOv5 architecture were found to be 0.667, 0.667 and 0.667, respectively. ConclusionsOur study reveals that OCSCC lesions carry discriminative visual appearances, which can be identified by deep learning algorithm. Artificial intelligence shows promise in the prediagnosis of oral cancer lesions. The success rates will increase in the training models of the data set that will be formed with more images.

Heart proteomic profiling discovers MYH6 and COX5B as biomarkers for sudden unexplained death

Sudden unexplained death (SUD) is not uncommon in forensic pathology. Yet, diagnosis of SUD remains challenging due to lack of specific biomarkers. This study aimed to screen differentially expressed proteins (DEPs) and validate their usefulness as diagnostic biomarkers for SUD cases. We designed a three-phase investigation, where in the discovery phase, formalin-fixed paraffin-embedded (FFPE) heart specimens were screened through label-free proteomic analysis of cases dying from SUD, mechanical injury and carbon monoxide (CO) intoxication. A total of 26 proteins were identified to be DEPs for the SUD cases after rigorous criterion. Bioinformatics and Adaboost-recursive feature elimination (RFE) analysis further revealed that three of the 26 proteins (MYH6, COX5B and TNNT2) were potential discriminative biomarkers. In the training phase, MYH6 and COX5B were verified to be true DEPs in cardiac tissues from 29 independent SUD cases as compared with a serial of control cases (n = 42). Receiver operating characteristic (ROC) analysis illustrated that combination of MYH6 and COX5B achieved optimal diagnostic sensitivity (89.7 %) and specificity (84.4 %), with area under the curve (AUC) being 0.91. A diagnostic software based on the logistic regression formula derived from the training phase was then constructed. In the validation phase, the diagnostic software was applied to eight authentic SUD cases, seven (87.5 %) of which were accurately recognized. Our study provides a valid strategy towards practical diagnosis of SUD by integrating cardiac MYH6 and COX5B as dual diagnostic biomarkers.

Facilitators of and Barriers to the Use of a Digital Self-Management Service for Diagnostic Testing: Focus Group Study With Potential Users.

Health care lags in digital transformation, despite the potential of technology to improve the well-being of individuals. The COVID-19 pandemic has accelerated the uptake of technology in health care and increased individuals' willingness to perform self-management using technology. A web-based service, Directlab Online, provides consumers with direct digital access to diagnostic test packages, which can digitally support the self-management of health. This study aims to identify the facilitators, barriers, and needs of Directlab Online, a self-management service for web-based access to diagnostic testing. A qualitative method was used from a potential user's perspective. The needs and future needs for, facilitators of, and barriers to the use of Directlab Online were evaluated. Semistructured focus group meetings were conducted in 2022. Two focus groups were focused on sexually transmitted infection test packages and 2 were focused on prevention test packages. Data analysis was performed according to the principles of the Framework Method. The Consolidated Framework for Implementation Research was used to categorize the facilitators and barriers. In total, 19 participants, with a mean age of 34.32 (SD 14.70) years, participated in the focus groups. Important barriers were a lack of privacy information, too much and difficult information, and a commercial appearance. Important facilitators were the right amount of information, the right kind of tests, and the involvement of a health care professional. The need for a service such as Directlab Online was to ensure its availability for users' health and to maintain their health. According to the participants, facilitators and barriers were comprehension of the information, the goal of the website, and the overall appearance of the service. Although the service was developed in cocreation with health care professionals and users, the needs did not align. The users preferred understandable and adequate, but not excessive, information. In addition, they preferred other types of tests to be available on the service. For future research, it would be beneficial to focus on cocreation between the involved medical professionals and users to develop, improve, and implement a service such as Directlab Online.

The plebeian Graph Library: A WebGL based network visualisation and diagnostics package

The plebeian Graph Library: A WebGL based network visualisation and diagnostics package

Implementation of a rapid diagnostic assay package for cryptococcosis, histoplasmosis and tuberculosis in people living with HIV in Paraguay

BackgroundOpportunistic infections (OIs) are common causes of mortality among people living with HIV (PLHIV). We determined prevalence and 30-day mortality due to histoplasmosis, cryptococcosis, and TB in PLHIV with advanced HIV disease (AHD).MethodsPLHIV 18 years and older, with a CD4 + T-cell count of less than 350 cells/mm3 newly diagnosed with HIV infection or re-engaged in care after being without ART for more than 90 days (Group A). The second group included symptomatic PLHIV regardless of ART status or CD4 + T-cell count (Group B); all followed for 30 days. Detection of Histoplasma Ag (HisAg) in urine was done by enzyme immunoassay (EIA), Cryptococcus antigen (CrAg) was detected in serum and cerebrospinal fluid (CSF) specimens by lateral flow assay (LFA), and lipoarabinomannan (LAM) detection in urine was by LFA (TB LAM) and in sputum by GeneXpert for diagnosis of Mycobacterium infections.ResultsFrom August 2021 to June 2022, 491 PLHIV were enrolled; 482 (98%) had a CD4 + T-cell result, and 381 patients (79%) were classified with AHD according to CD4 + T-cell count (< 200 CD4/mm3). Frequency of an OI was 38% (n = 145/381). Antigen test positivity rate was 16% (72/467) for TB-LAM, 9% (43/464) for HisAg, and 11% (51/484) for CrAg. Twenty-one of 34 (62%) patients receiving CSF CrAg tests were positive, confirming meningitis. Significant differences in 30-day mortality were observed in patients with an OI (16%) vs. no OI (7%) (p = 0.002). Mortality was highest in patients with histoplasmosis (25%), co-infection (22%), cryptococcosis (18% overall; 19% for cryptococcal meningitis), and TB (10%).ConclusionsTB and fungal OIs, including co-infection, were common in PLHIV in Paraguay and had high associated mortality. Laboratories and health facilities need access to CD4 + T-cell testing and rapid diagnostic assays.

Application of artificial intelligence-assisted image diagnosis software based on volume data reconstruction technique in medical imaging practice teaching

BackgroundIn medical imaging courses, due to the complexity of anatomical relationships, limited number of practical course hours and instructors, how to improve the teaching quality of practical skills and self-directed learning ability has always been a challenge for higher medical education. Artificial intelligence-assisted diagnostic (AISD) software based on volume data reconstruction (VDR) technique is gradually entering radiology. It converts two-dimensional images into three-dimensional images, and AI can assist in image diagnosis. However, the application of artificial intelligence in medical education is still in its early stages. The purpose of this study is to explore the application value of AISD software based on VDR technique in medical imaging practical teaching, and to provide a basis for improving medical imaging practical teaching.MethodsTotally 41 students majoring in clinical medicine in 2017 were enrolled as the experiment group. AISD software based on VDR was used in practical teaching of medical imaging to display 3D images and mark lesions with AISD. Then annotations were provided and diagnostic suggestions were given. Also 43 students majoring in clinical medicine from 2016 were chosen as the control group, who were taught with the conventional film and multimedia teaching methods. The exam results and evaluation scales were compared statistically between groups.ResultsThe total skill scores of the test group were significantly higher compared with the control group (84.51 ± 3.81 vs. 80.67 ± 5.43). The scores of computed tomography (CT) diagnosis (49.93 ± 3.59 vs. 46.60 ± 4.89) and magnetic resonance (MR) diagnosis (17.41 ± 1.00 vs. 16.93 ± 1.14) of the experiment group were both significantly higher. The scores of academic self-efficacy (82.17 ± 4.67) and self-directed learning ability (235.56 ± 13.50) of the group were significantly higher compared with the control group (78.93 ± 6.29, 226.35 ± 13.90).ConclusionsApplying AISD software based on VDR to medical imaging practice teaching can enable students to timely obtain AI annotated lesion information and 3D images, which may help improve their image reading skills and enhance their academic self-efficacy and self-directed learning abilities.

Does a new case-based payment system promote the construction of the ordered health delivery system? Evidence from a pilot city in China

BackgroundThe construction of the ordered health delivery system in China aims to enhance equity and optimize the efficient use of medical resources by rationally allocating patients to different levels of medical institutions based on the severity of their condition. However, superior hospitals have been overcrowded, and primary healthcare facilities have been underutilized in recent years. China has developed a new case-based payment method called “Diagnostic Intervention Package” (DIP). The government is trying to use this economic lever to encourage medical institutions to actively assume treatment tasks consistent with their functional positioning and service capabilities.MethodsThis study takes Tai’an, a DIP pilot city, as a case study and uses an interrupted time series analysis to analyze the impact of DIP reform on the case severity and service scope of medical institutions at different levels.ResultsThe results show that after the DIP reform, the proportion of patients receiving complicated procedures (tertiary hospitals: β3 = 0.197, P < 0.001; secondary hospitals: β3 = 0.132, P = 0.020) and the case mix index (tertiary hospitals: β3 = 0.022, P < 0.001; secondary hospitals: β3 = 0.008, P < 0.001) in tertiary and secondary hospitals increased, and the proportion of primary-DIP-groups cases decreased (tertiary hospitals: β3 = -0.290, P < 0.001; secondary hospitals: β3 = -1.200, P < 0.001), aligning with the anticipated policy objectives. However, the proportion of patients receiving complicated procedures (β3 = 0.186, P = 0.002) and the case mix index (β3 = 0.002, P < 0.001) in primary healthcare facilities increased after the reform, while the proportion of primary-DIP-groups cases (β3 = -0.515, P = 0.005) and primary-DIP-groups coverage (β3 = -2.011, P < 0.001) decreased, which will reduce the utilization efficiency of medical resources and increase inequity.ConclusionThe DIP reform did not effectively promote the construction of the ordered health delivery system. Policymakers need to adjust economic incentives and implement restraint mechanisms to regulate the behavior of medical institutions.

Unlocking the Potential: Investigating Dental Practitioners' Willingness to Embrace Artificial Intelligence in Dental Practice.

Artificial intelligence (AI) holds significant promise for transforming healthcare delivery, including dentistry. However, the successful integration of AI into dental practice necessitates an understanding of dental professionals' perspectives, attitudes, and readiness to adopt AI technology. This study aimed to explore dental professionals' perceptions, attitudes, and practices regarding AI adoption in dentistry. This cross-sectional study was conducted among 256 dental professionals using an online questionnaire. Participants were assessed for familiarity with AI technology, perceived barriers to adoption, attitudes towards AI, current usage patterns, and factors influencing adoption decisions. Data are analysed using descriptive statistics, including frequencies, percentages, means, and standard deviations. Inferential statistics, such as chi-square tests and regression analysis, were employed to examine associations between variables and identify predictors of AI adoption in dentistry. The study surveyed 256 dental professionals from various regions across India, primarily aged 30 to 50 years (mean age: 42.6), with a nearly equal gender split (male: 48.4%, female: 51.6%) and high educational attainment (67.8% with master's or doctoral degrees). Private practices were predominant (56.3%). The diagnostic algorithms and treatment planning software were well known (77.3% and 70.3% familiarity, respectively). Technical concerns (average score: 3.82 ± 0.68) were the main barriers to AI adoption, followed by financial considerations (average score: 3.45 ± 0.72), ethical and legal issues (average score: 3.21 ± 0.65), and organizational factors (average score: 3.67 ± 0.71). Despite these concerns, most participants had positive attitudes towards AI (70.3% agreed). Current usage varied, with diagnostic support and administrative tasks being the most common (44.5% and 82.8% usage, respectively). Perceived utility (average score: 4.12 ± 0.75) and ease of use (average score: 3.98 ± 0.69) significantly influenced adoption, as identified by regression analysis (perceived utility: β = 0.342, p < 0.001; ease of use: β = 0.267, p = 0.005). This study provides valuable insights into AI adoption in dentistry, highlighting the multifaceted nature of barriers and facilitators that influence dental professionals' adoption decisions. Strategies to promote AI adoption should address practical considerations, ethical concerns, and educational needs to facilitate the integration of AI technology into dental practices.

Automatic early detection of induced colic in horses using accelerometer devices.

To seek appropriate veterinary attention for horses with colic, owners must recognise early signs. Direct observation of horse behaviour has several drawbacks: it is time-consuming, hard to see subtle and common behavioural signs, and is based on intuition and subjective decisions. Due to recent advances in wearables and artificial intelligence, it may be possible to develop diagnostic software that can automatically detect colic signs. To develop a software algorithm to aid in the detection of colic signs and levels of pain. In vivo experiments. Transient colic was induced in eight experimental mares with luteolytic doses of prostaglandin. Veterinarians observed the horses before and throughout the interventions and assigned pain scores which were used to separate colic episodes into none (pain score ≤5), level 1 (pain score 6-10) or level 2 (pain score ≥11). Accelerometric data and videos were collected throughout the experiments and using accelerometric data, the horse's behaviour was classified into normal and 10 pain-related behaviours and an activity index was calculated. Models were designed that utilised behaviour and activity index characteristics both detecting the presence of colic and assessing its severity. To determine the accuracy of the model, the ground truth, that is the veterinarians' observation of colic signs and assessment of pain level, was compared with the automatic detection system. The cross-validation analysis demonstrated an accuracy of 91.2% for detecting colic and an accuracy of 93.8% in differentiating between level 1 colic and level 2 colic. The model was able to accurately classify 10 pain-related behaviours and distinguish them from normal behaviour with a high accuracy. We included a limited number of horses with severe pain related behaviours in the dataset. This constraint affects the accuracy of categorising colic severity rather than limiting the algorithms' capacity to identify early colic signs. Our system for early detection of colic in horses is unique and innovative, and it can distinguish between colic of varying severity.

Multi-Convolutional Neural Network-Based Diagnostic Software for the Presumptive Determination of Non-Dermatophyte Molds

Based on the literature data, the incidence of superficial and invasive non-dermatophyte mold infection (NDMI) has increased. Many of these infections are undiagnosed or misdiagnosed, thus causing inadequate treatment procedures followed by critical conditions or even mortality of the patients. Accurate diagnosis of these infections requires complex mycological analyses and operator skills, but simple, fast, and more efficient mycological tests are still required to overcome the limitations of conventional fungal diagnostic procedures. In this study, software has been developed to provide an efficient mycological diagnosis using a trained convolutional neural network (CNN) model as a core classifier. Using EfficientNet-B2 architecture and permanent slides of NDM isolated from patient’s materials (personal archive of Prof. Otašević, Department of Microbiology and Immunology, Medical Faculty, University of Niš, Serbia), a multi-CNN model has been trained and then integrated into the diagnostic tool, with a 93.73% accuracy of the main model. The Grad-CAM visualization model has been used for further validation of the pattern recognition of the model. The software, which makes the final diagnosis based on the rule of the major method, has been tested with images provided by different European laboratories, showing an almost faultless accuracy with different test images.

Evaluation of artificial intelligence for detecting periapical lesions on panoramic radiographs

Background/Aim: The aim of this study is to evaluate the function of diagnostic computer software designed for the detection of periapical lesions on panoramic images with deep learning methods. Material and Methods: In our study, more than 500 adult retrospective panoramic radiography images obtained randomly were evaluated, and periapical lesions were labeled on the radiographs with the ResultLab.Ai labeling program (ResultLab.AI, Istanbul, Turkey). All images have been rechecked and verified by Oral, Dental and Maxillofacial Radiology experts. The model used is a U-Net based architecture customized for dental radiographs and optimized for fast operation. What is meant by customization is the structures called "Attention Gate" added to the architecture to draw the model's attention to the lesions. Results: Mask estimation was made separately for each image and F1 and IoU scores were calculated by comparing them with the marked masks. A list was created from the calculated F1 and IoU scores. The F1 and IoU score of the entire data set was calculated by taking the average of the values in this list. In IoU score calculation, Keras library's Mean IoU metric was used to calculate for 2 classes. In the F1 score calculation, the F1 score metric of the SKLearn library was used to calculate the unweighted average for 2 classes. As a result, the IoU-Score was obtained as 0.8578 and the F1-Score as 0.8587. Conclusions: Detection of periapical lesions on panoramic radiographs with an artificial intelligence approach will help practitioners reach definitive conclusions even in lesions that are likely to be overlooked. In addition, success rates for deep learning methods improve as data set size grows. As the number of images rises, the training models' success rates will follow.

IMPROVEMENT OF THE METHOD OF DIAGNOSTICS OF HIGH-VOLTAGE TRANSFORMERS VOLTAGE BASED ON THE RESULTS OF CHROMATOGRAPHIC ANALYSES OF TRANSFORMER OILS

The relevance of the work is due to the need to improve methods for diagnosing high-voltage oil-filled voltage transformers with paper-oil electrical insulation, which makes it possible to solve the problems of monitoring the technical condition, finding the place and determining the causes of failure (malfunction), and predicting the technical condition of this equipment on the basis of appropriate diagnostic software. The aim of the work is to ensure the reliability of operation of high-voltage oil-filled voltage transformers with paper-oil electrical insulation by improving the method of their diagnosis based on the results of analyses of samples of mineral transformer oils by gas chromatography. In accordance with the research objective, the following tasks were solved: 1. Determine the content of such dissolved diagnostic components in samples of mineral transformer oils from high-voltage oil-filled voltage transformers of NKF-110, NKF-220, NKF-330 types as gases H2, CH4, C2H6, C2H4, C2H2, CO, CO2, O2, N2; additive "Ionol"; furan compounds (2-furyl alcohol; 2-furfural; 2-acetylfuran; 5-methyl-2-furfural) and the sum of their concentrations. 2. The basic procedures for diagnosing high-voltage oil-filled voltage transformers based on the results of analyses of mineral transformer oil samples with the determination of the content of dissolved diagnostic components in them were improved. For the diagnosis of high-voltage oil-filled voltage transformers based on the results of gas chromatographic analyses of samples of mineral transformer oils from this equipment, it is recommended to use the following methods: "Limit level of concentrations", "Tabular method", "Graphic images of defects", "ETRA", "Duval triangle", "Visual inspection after disassembly of high-voltage oil-filled voltage transformers". To determine the degree of danger of the development of the identified probable defects, the method of determining the "Relative growth rate of dissolved gas concentrations" can be used. The prospect of practical application of the results of the work is to ensure the reliability of operation of high-voltage oil-filled voltage transformers by improving the method of their diagnosis based on the results of analyses of samples of mineral transformer oils with simultaneous determination of the content of dissolved gases, the additive "Ionol", furan compounds by gas chromatography.

Искусственный интеллект в здравоохранении России: сбор и подготовка данных для машинного обучения.

This scientific article draws attention to the importance of collecting and preparing high-quality medical data for the development of artificial intelligence (AI) systems in Russian healthcare. Materials and methods. The main emphasis is on the need to create a unified federal standard for collecting structured digitized medical data, which will unify the processes of collecting and generating datasets for medical organizations. Results. The article proposes the introduction of digital assistants, including pre-medical diagnostic questionnaires, mobile applications and software modules for medical examinations, which will reduce the time spent filling out medical documentation and ensure the collection of more complete and accurate information. The next step is to reengineer the health data collection process, including patient engagement via smartphones and personal health assistants. The article identifies methodological problems such as unreliability, incompleteness and lack of extensional knowledge in the collected data, and proposes a roadmap for the development of a health data collection system, including the creation of digital standards, data collection methods and the development of AI-based decision support. Conclusions. Solving the described problems and the data development plan are important for the successful implementation of the national project «Data Economy», supporting the development and application of AI systems in the medical field.