Improved accuracy in colorectal cancer tissue decomposition through refinement of established deep learning solutions.

Hematoxylin and eosin-stained biopsy slides are regularly available for colorectal cancer patients. These slides are often not used to define objective biomarkers for patient stratification and treatment selection. Standard biomarkers often pertain to costly and slow genetic tests. However, recent work has shown that relevant biomarkers can be extracted from these images using convolutional neural networks (CNNs). The CNN-based biomarkers predicted colorectal cancer patient outcomes comparably to gold standards. Extracting CNN-biomarkers is fast, automatic, and of minimal cost. CNN-based biomarkers rely on the ability of CNNs to recognize distinct tissue types from microscope whole slide images. The quality of these biomarkers (coined 'Deep Stroma') depends on the accuracy of CNNs in decomposing all relevant tissue classes. Improving tissue decomposition accuracy is essential for improving the prognostic potential of CNN-biomarkers. In this study, we implemented a novel training strategy to refine an established CNN model, which then surpassed all previous solutions . We obtained a 95.6% average accuracy in the external test set and 99.5% in the internal test set. Our approach reduced errors in biomarker-relevant classes, such as Lymphocytes, and was the first to include interpretability methods. These methods were used to better apprehend our model's limitations and capabilities.

Domain-specific transfer learning in the automated scoring of tumor-stroma ratio from histopathological images of colorectal cancer.

Tumor-stroma ratio (TSR) is a prognostic factor for many types of solid tumors. In this study, we propose a method for automated estimation of TSR from histopathological images of colorectal cancer. The method is based on convolutional neural networks which were trained to classify colorectal cancer tissue in hematoxylin-eosin stained samples into three classes: stroma, tumor and other. The models were trained using a data set that consists of 1343 whole slide images. Three different training setups were applied with a transfer learning approach using domain-specific data i.e. an external colorectal cancer histopathological data set. The three most accurate models were chosen as a classifier, TSR values were predicted and the results were compared to a visual TSR estimation made by a pathologist. The results suggest that classification accuracy does not improve when domain-specific data are used in the pre-training of the convolutional neural network models in the task at hand. Classification accuracy for stroma, tumor and other reached 96.1% on an independent test set. Among the three classes the best model gained the highest accuracy (99.3%) for class tumor. When TSR was predicted with the best model, the correlation between the predicted values and values estimated by an experienced pathologist was 0.57. Further research is needed to study associations between computationally predicted TSR values and other clinicopathological factors of colorectal cancer and the overall survival of the patients.

Data Analytics in Healthcare: A Tertiary Study.

The field of healthcare has seen a rapid increase in the applications of data analytics during the last decades. By utilizing different data analytic solutions, healthcare areas such as medical image analysis, disease recognition, outbreak monitoring, and clinical decision support have been automated to various degrees. Consequently, the intersection of healthcare and data analytics has received scientific attention to the point of numerous secondary studies. We analyze studies on healthcare data analytics, and provide a wide overview of the subject. This is a tertiary study, i.e., a systematic review of systematic reviews. We identified 45 systematic secondary studies on data analytics applications in different healthcare sectors, including diagnosis and disease profiling, diabetes, Alzheimer's disease, and sepsis. Machine learning and data mining were the most widely used data analytics techniques in healthcare applications, with a rising trend in popularity. Healthcare data analytics studies often utilize four popular databases in their primary study search, typically select 25-100 primary studies, and the use of research guidelines such as PRISMA is growing. The results may help both data analytics and healthcare researchers towards relevant and timely literature reviews and systematic mappings, and consequently, towards respective empirical studies. In addition, the meta-analysis presents a high-level perspective on prominent data analytics applications in healthcare, indicating the most popular topics in the intersection of data analytics and healthcare, and provides a big picture on a topic that has seen dozens of secondary studies in the last 2 decades.

DeepFake knee osteoarthritis X-rays from generative adversarial neural networks deceive medical experts and offer augmentation potential to automatic classification.

Recent developments in deep learning have impacted medical science. However, new privacy issues and regulatory frameworks have hindered medical data sharing and collection. Deep learning is a very data-intensive process for which such regulatory limitations limit the potential for new breakthroughs and collaborations. However, generating medically accurate synthetic data can alleviate privacy issues and potentially augment deep learning pipelines. This study presents generative adversarial neural networks capable of generating realistic images of knee joint X-rays with varying osteoarthritis severity. We offer 320,000 synthetic (DeepFake) X-ray images from training with 5,556 real images. We validated our models regarding medical accuracy with 15 medical experts and for augmentation effects with an osteoarthritis severity classification task. We devised a survey of 30 real and 30 DeepFake images for medical experts. The result showed that on average, more DeepFakes were mistaken for real than the reverse. The result signified sufficient DeepFake realism for deceiving the medical experts. Finally, our DeepFakes improved classification accuracy in an osteoarthritis severity classification task with scarce real data and transfer learning. In addition, in the same classification task, we replaced all real training data with DeepFakes and suffered only a [Formula: see text] loss from baseline accuracy in classifying real osteoarthritis X-rays.

Cross-cultural adaptation and validation of the Kerlan-Jobe Orthopaedic Clinic shoulder and elbow score in Finnish-speaking overhead athletes.

BACKGROUND: The Kerlan-Jobe Orthopaedic Clinic Shoulder and Elbow score (KJOC) is developed to evaluate the shoulder and elbow function in overhead athletes. To date, the score has not been adapted into Finnish language. The aim of this study was to perform a cross-cultural adaptation of the Kerlan-Jobe Orthopaedic Clinic Shoulder and Elbow score (KJOC) into Finnish language and evaluate its validity, reliability, and responsiveness in overhead athletes. METHODS: Forward-backward translation method was followed in the cross-cultural adaptation process. Subsequently, 114 overhead athletes (52 males, 62 females, mean age 18.1 ± 2.8 years) completed the Finnish version of KJOC score, Disabilities of the Arm, Shoulder and Hand (DASH), American Shoulder and Elbow Surgeons Standardized Shoulder Assessment Form (ASES) and RAND-36 to assess validity of the KJOC score. To evaluate reliability and responsiveness, the participants filled in the KJOC score 16 days and eight months after the first data collection. Validity, reliability, and responsiveness of the Finnish KJOC score were statistically tested. RESULTS: Minor modifications were made during the cross-cultural translation and adaptation process, which were related to culture specific terminology in sports and agreed by an expert committee. Construct validity of the KJOC score was moderate to high, based on the correlations with DASH (r = - 0.757); DASH sports module (r = - 0.667); ASES (r = 0.559); and RAND-36 (r = 0.397) questionnaires. Finnish KJOC score showed excellent internal consistency (α = 0.92) and good test-retest reliability (2-way mixed-effects model ICC = 0.77) with acceptable measurement error level (SEM 5.5; MDC 15.1). Ceiling effect was detected for asymptomatic athletes in each item (23.2-61.1%), and for symptomatic athletes in item 5 (47.4%). Responsiveness of the Finnish KJOC score could not be confirmed due to conflicting follow-up results. CONCLUSION: The Finnish KJOC score was found to be a valid and reliable questionnaire measuring the self-reported upper arm status in Finnish-speaking overhead athletes.

An Automatic Method for Assessing Spiking of Tibial Tubercles Associated with Knee Osteoarthritis.

Efficient and scalable early diagnostic methods for knee osteoarthritis are desired due to the disease's prevalence. The current automatic methods for detecting osteoarthritis using plain radiographs struggle to identify the subjects with early-stage disease. Tibial spiking has been hypothesized as a feature of early knee osteoarthritis. Previous research has demonstrated an association between knee osteoarthritis and tibial spiking, but the connection to the early-stage disease has not been investigated. We study tibial spiking as a feature of early knee osteoarthritis. Additionally, we develop a deep learning based model for detecting tibial spiking from plain radiographs. We collected and graded 913 knee radiographs for tibial spiking. We conducted two experiments: experiments A and B. In experiment A, we compared the subjects with and without tibial spiking using Mann-Whitney U-test. Experiment B consisted of developing and validating an interpretative deep learning based method for predicting tibial spiking. The subjects with tibial spiking had more severe Kellgren-Lawrence grade, medial joint space narrowing, and osteophyte score in the lateral tibial compartment. The developed method achieved an accuracy of 0.869. We find tibial spiking a promising feature in knee osteoarthritis diagnosis. Furthermore, the detection can be automatized.

Predicting ACL Injury Using Machine Learning on Data From an Extensive Screening Test Battery of 880 Female Elite Athletes.

BACKGROUND: Injury risk prediction is an emerging field in which more research is needed to recognize the best practices for accurate injury risk assessment. Important issues related to predictive machine learning need to be considered, for example, to avoid overinterpreting the observed prediction performance. PURPOSE: To carefully investigate the predictive potential of multiple predictive machine learning methods on a large set of risk factor data for anterior cruciate ligament (ACL) injury; the proposed approach takes into account the effect of chance and random variations in prediction performance. STUDY DESIGN: Case-control study; Level of evidence, 3. METHODS: The authors used 3-dimensional motion analysis and physical data collected from 791 female elite handball and soccer players. Four common classifiers were used to predict ACL injuries (n = 60). Area under the receiver operating characteristic curve (AUC-ROC) averaged across 100 cross-validation runs (mean AUC-ROC) was used as a performance metric. Results were confirmed with repeated permutation tests (paired Wilcoxon signed-rank-test; P < .05). Additionally, the effect of the most common class imbalance handling techniques was evaluated. RESULTS: For the best classifier (linear support vector machine), the mean AUC-ROC was 0.63. Regardless of the classifier, the results were significantly better than chance, confirming the predictive ability of the data and methods used. AUC-ROC values varied substantially across repetitions and methods (0.51-0.69). Class imbalance handling did not improve the results. CONCLUSION: The authors' approach and data showed statistically significant predictive ability, indicating that there exists information in this prospective data set that may be valuable for understanding injury causation. However, the predictive ability remained low from the perspective of clinical assessment, suggesting that included variables cannot be used for ACL prediction in practice.

Change of Direction Biomechanics in a 180-Degree Pivot Turn and the Risk for Noncontact Knee Injuries in Youth Basketball and Floorball Players.

BACKGROUND: Studies investigating biomechanical risk factors for knee injuries in sport-specific tasks are needed. PURPOSE: To investigate the association between change of direction (COD) biomechanics in a 180-degree pivot turn and knee injury risk among youth team sport players. STUDY DESIGN: Cohort study; Level of evidence, 2. METHODS: A total of 258 female and male basketball and floorball players (age range, 12-21 years) participated in the baseline COD test and follow-up. Complete data were obtained from 489 player-legs. Injuries, practice, and game exposure were registered for 12 months. The COD test consisted of a quick ball pass before and after a high-speed 180-degree pivot turn on the force plates. The following variables were analyzed: peak vertical ground-reaction force (N/kg); peak trunk lateral flexion angle (degree); peak knee flexion angle (degree); peak knee valgus angle (degree); peak knee flexion moment (N·m/kg); peak knee abduction moment (N·m/kg); and peak knee internal and external rotation moments (N·m/kg). Legs were analyzed separately and the mean of 3 trials was used in the analysis. Main outcome measure was a new acute noncontact knee injury. RESULTS: A total of 18 new noncontact knee injuries were registered (0.3 injuries/1000 hours of exposure). Female players sustained 14 knee injuries and male players 4. A higher rate of knee injuries was observed in female players compared with male players (incidence rate ratio, 6.2; 95% CI, 2.1-21.7). Of all knee injuries, 8 were anterior cruciate ligament (ACL) injuries, all in female players. Female players displayed significantly larger peak knee valgus angles compared with male players (mean for female and male players, respectively: 13.9°± 9.4° and 2.0°± 8.5°). No significant associations between biomechanical variables and knee injury risk were found. CONCLUSION: Female players were at increased risk of knee and ACL injury compared with male players. Female players performed the 180-degree pivot turn with significantly larger knee valgus compared with male players. However, none of the investigated variables was associated with knee injury risk in youth basketball and floorball players.

Precision exercise medicine: predicting unfavourable status and development in the 20-m shuttle run test performance in adolescence with machine learning.

OBJECTIVES: To assess the ability to predict individual unfavourable future status and development in the 20m shuttle run test (20MSRT) during adolescence with machine learning (random forest (RF) classifier). METHODS: Data from a 2-year observational study (2013‒2015, 12.4±1.3 years, n=633, 50% girls), with 48 baseline characteristics (questionnaires (demographics, physical, psychological, social and lifestyle factors), objective measurements (anthropometrics, fitness characteristics, physical activity, body composition and academic scores)) were used to predict: (Task 1) unfavourable future 20MSRT status (identification of individuals in the lowest 20MSRT tertile after 2 years), and (Task 2) unfavourable 20MSRT development (identification of individuals with 20MSRT development in the lowest tertile among adolescents with baseline 20MSRT below median level). RESULTS: Prediction performance for future 20MSRT status (Task 1) was (area under the receiver operating characteristic curve, AUC) 83% and 76%, sensitivity 80% and 60%, and specificity 78% and 79% in girls and boys, respectively. Twenty variables showed predictive power in boys, 14 in girls, including fitness characteristics, physical activity, academic scores, adiposity, life enjoyment, parental support, social status in school and perceived fitness.Prediction performance for future development (Task 2) was lower and differed statistically from random level only in girls (AUC 68% and 40% in girls and boys). CONCLUSION: RF classifier predicted future unfavourable status in 20MSRT and identified potential individuals for interventions based on a holistic profile (14‒20 baseline characteristics). The MATLAB script and functions employing the RF classifier of this study are available for future precision exercise medicine research.

The standing knee lift test is not a useful screening tool for time loss from low back pain in youth basketball and floorball players.

OBJECTIVES: The aim of this study was to investigate the association between pelvic kinematics during the standing knee lift (SKL) test and low back pain (LBP) in youth floorball and basketball players. DESIGN: A prospective cohort study. SETTING: Finnish elite youth floorball and basketball players. PARTICIPANTS: Finnish elite youth female and male floorball and basketball players (n = 258, mean age 15.7 ± 1.8). MAIN OUTCOME MEASURES: LBP resulting in time loss from practice and games was recorded over a 12-month period and verified by a study physician. Associations between LBP and sagittal plane pelvic tilt and frontal plane pelvic obliquity during the SKL test as measured at baseline were investigated. Individual training and game hours were recorded, and Cox's proportional hazard models with mixed effects were used for the analysis. RESULTS: Cox analyses revealed that sagittal plane pelvic tilt and frontal plane pelvic obliquity were not associated with LBP in floorball and basketball players during the follow-up. The hazard ratios for pelvic tilt and pelvic obliquity ranged between 0.93 and 1.08 (95% CIs between 0.91 and 1.07 and 0.83 and 1.29), respectively. CONCLUSIONS: Pelvic movement during the SKL test is not associated with future LBP in youth floorball and basketball players.

New Machine Learning Approach for Detection of Injury Risk Factors in Young Team Sport Athletes.

The purpose of this article is to present how predictive machine learning methods can be utilized for detecting sport injury risk factors in a data-driven manner. The approach can be used for finding new hypotheses for risk factors and confirming the predictive power of previously recognized ones. We used three-dimensional motion analysis and physical data from 314 young basketball and floorball players (48.4% males, 15.72±1.79 yr, 173.34±9.14 cm, 64.65±10.4 kg). Both linear (L1-regularized logistic regression) and non-linear methods (random forest) were used to predict moderate and severe knee and ankle injuries (N=57) during three-year follow-up. Results were confirmed with permutation tests and predictive risk factors detected with Wilcoxon signed-rank-test (p<0.01). Random forest suggested twelve consistent injury predictors and logistic regression twenty. Ten of these were suggested in both models; sex, body mass index, hamstring flexibility, knee joint laxity, medial knee displacement, height, ankle plantar flexion at initial contact, leg press one-repetition max, and knee valgus at initial contact. Cross-validated areas under receiver operating characteristic curve were 0.65 (logistic regression) and 0.63 (random forest). The results highlight the difficulty of predicting future injuries, but also show that even with models having relatively low predictive power, certain predictive injury risk factors can be consistently detected.

Deep learning approach for prediction of impact peak appearance at ground reaction force signal of running activity.

Protruding impact peak is one of the features of vertical ground reaction force (GRF) that is related to injury risk while running. The present research is dedicated to predicting GRF impact peak appearance by setting a binary classification problem. Kinematic data, namely a number of raw signals in the sagittal plane, collected by the Vicon motion capture system (Oxford Metrics Group, UK) were employed as predictors. Therefore, the input data for the predictive model are presented as a multi-channel time series. Deep learning techniques, namely five convolutional neural network (CNN) models were applied to the binary classification analysis, based on a Multi-Layer Perceptron (MLP) classifier, support vector machine (SVM), logistic regression, k-nearest neighbors (kNN), and random forest algorithms. SVM, logistic regression, and random forest classifiers demonstrated performances that do not statistically significantly differ. The best classification accuracy achieved is 81.09% ± 2.58%. Due to good performance of the models, this study serves as groundwork for further application of deep learning approaches to predicting kinetic information based on this kind of input data.

Erratum: Multilabel segmentation of cancer cell culture on vascular structures with deep neural networks (Erratum).

[This corrects the article DOI: 10.1117/1.JMI.7.2.024001.].

Multilabel segmentation of cancer cell culture on vascular structures with deep neural networks.

New increasingly complex in vitro cancer cell models are being developed. These new models seem to represent the cell behavior in vivo more accurately and have better physiological relevance than prior models. An efficient testing method for selecting the most optimal drug treatment does not exist to date. One proposed solution to the problem involves isolation of cancer cells from the patients' cancer tissue, after which they are exposed to potential drugs alone or in combinations to find the most optimal medication. To achieve this goal, methods that can efficiently quantify and analyze changes in tested cell are needed. Our study aimed to detect and segment cells and structures from cancer cell cultures grown on vascular structures in phase-contrast microscope images using U-Net neural networks to enable future drug efficacy assessments. We cultivated prostate carcinoma cell lines PC3 and LNCaP on the top of a matrix containing vascular structures. The cells were imaged with a Cell-IQ phase-contrast microscope. Automatic analysis of microscope images could assess the efficacy of tested drugs. The dataset included 36 RGB images and ground-truth segmentations with mutually not exclusive classes. The used method could distinguish vascular structures, cells, spheroids, and cell matter around spheroids in the test images. Some invasive spikes were also detected, but the method could not distinguish the invasive cells in the test images.

Altered hip control during a standing knee-lift test is associated with increased risk of knee injuries.

A few prospective studies have investigated hip and pelvic control as a risk factor for lower extremity (LE) injuries. The purpose of this study was to investigate whether deficits in hip and lumbopelvic control during standing knee-lift test are associated with increased risk of acute knee and LE injuries in youth team sports. At baseline, 258 basketball and floorball players (aged 12-21 years) participated in a standing knee-lift test using 3-dimensional motion analysis. Two trials per leg were recorded from each participant. Peak sagittal plane pelvic tilt and frontal plane pelvic drop/hike were measured. Both continuous and categorical variables were analyzed. New non-contact LE injuries, and match and training exposure, were recorded for 12 months. Seventy acute LE injuries were registered. Of these, 17 were knee injuries (eight ACL ruptures) and 35 ankle injuries. Risk factor analyses showed that increased contralateral pelvic hike was significantly associated with knee injury risk when using categorical variable (HR for high vs low group 4.07; 95% CI 1.32-12.6). Furthermore, significant association was found between high lateral pelvic hike angles and ACL injury risk in female players (HR for high vs low group 9.10; 95% CI 1.10-75.2). Poor combined sensitivity and specificity of the test was observed. In conclusion, increased contralateral pelvic hike is associated with non-contact knee injury risk among young team sport players and non-contact ACL injuries among female players. More research to determine the role of pelvic control as a risk factor for knee injuries is needed.

A hierarchical cluster analysis to determine whether injured runners exhibit similar kinematic gait patterns.

Previous studies have suggested that runners can be subgrouped based on homogeneous gait patterns; however, no previous study has assessed the presence of such subgroups in a population of individuals across a wide variety of injuries. Therefore, the purpose of this study was to assess whether distinct subgroups with homogeneous running patterns can be identified among a large group of injured and healthy runners and whether identified subgroups are associated with specific injury location. Three-dimensional kinematic data from 291 injured and healthy runners, representing both sexes and a wide range of ages (10-66 years), were clustered using hierarchical cluster analysis. Cluster analysis revealed five distinct subgroups from the data. Kinematic differences between the subgroups were compared using one-way analysis of variance (ANOVA). Against our hypothesis, runners with the same injury types did not cluster together, but the distribution of different injuries within subgroups was similar across the entire sample. These results suggest that homogeneous gait patterns exist independent of injury location and that it is important to consider these underlying patterns when planning injury prevention or rehabilitation strategies.

Predicting hospital associated disability from imbalanced data using supervised learning.

Hospitalization of elderly patients can lead to serious adverse effects on their functional capability. Identifying the underlying factors leading to such adverse effects is an active area of medical research. The purpose of the current paper is to show the potential of artificial intelligence in the form of machine learning to complement the existing medical research. This is accomplished by studying the outcome of hospitalization of elderly patients as a supervised learning task. A rich set of features characterizing the medical and social situation of elderly patients is leveraged and using confusion matrices, association rule mining, and two different classes of supervised learning algorithms, it is shown that the need for help and supervision are the most important features predicting whether these patients will return home after hospitalization. Such findings can help to improve hospitalization and rehabilitation of elderly patients.

Balance Perturbations as a Measurement Tool for Trunk Impairment in Cross-Country Sit Skiing.

In cross-country sit-skiing, the trunk plays a crucial role in propulsion generation and balance maintenance. Trunk stability is evaluated by automatic responses to unpredictable perturbations; however, electromyography is challenging. The aim of this study was to identify a measure to group sit-skiers according to their ability to control the trunk. Seated in their competitive sit-ski, 10 male and 5 female Paralympic sit-skiers received 6 forward and 6 backward unpredictable perturbations in random order. k-means clustered trunk position at rest, delay to invert the trunk motion, and trunk range of motion significantly into 2 groups. In conclusion, unpredictable perturbations might quantify trunk impairment and may become an important tool in the development of an evidence-based classification system for cross-country sit-skiers.

Stiff Landings Are Associated With Increased ACL Injury Risk in Young Female Basketball and Floorball Players.

BACKGROUND: Few prospective studies have investigated the biomechanical risk factors of anterior cruciate ligament (ACL) injury. PURPOSE: To investigate the relationship between biomechanical characteristics of vertical drop jump (VDJ) performance and the risk of ACL injury in young female basketball and floorball players. STUDY DESIGN: Cohort study; Level of evidence, 3. METHODS: At baseline, a total of 171 female basketball and floorball players (age range, 12-21 years) participated in a VDJ test using 3-dimensional motion analysis. The following biomechanical variables were analyzed: (1) knee valgus angle at initial contact (IC), (2) peak knee abduction moment, (3) knee flexion angle at IC, (4) peak knee flexion angle, (5) peak vertical ground-reaction force (vGRF), and (6) medial knee displacement. All new ACL injuries, as well as match and training exposure, were then recorded for 1 to 3 years. Cox regression models were used to calculate hazard ratios (HRs) and 95% CIs. RESULTS: Fifteen new ACL injuries occurred during the study period (0.2 injuries/1000 player-hours). Of the 6 factors considered, lower peak knee flexion angle (HR for each 10° increase in knee flexion angle, 0.55; 95% CI, 0.34-0.88) and higher peak vGRF (HR for each 100-N increase in vGRF, 1.26; 95% CI, 1.09-1.45) were the only factors associated with increased risk of ACL injury. A receiver operating characteristic (ROC) curve analysis showed an area under the curve of 0.6 for peak knee flexion and 0.7 for vGRF, indicating a failed-to-fair combined sensitivity and specificity of the test. CONCLUSIONS: Stiff landings, with less knee flexion and greater vGRF, in a VDJ test were associated with increased risk of ACL injury among young female basketball and floorball players. However, although 2 factors (decreased peak knee flexion and increased vGRF) had significant associations with ACL injury risk, the ROC curve analyses revealed that these variables cannot be used for screening of athletes.