Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Pathology Pathology

SC-MIL: Supervised Contrastive Multiple Instance Learning for Imbalanced Classification in Pathology

ArXiv Preprint

Multiple Instance learning (MIL) models have been extensively used in pathology to predict biomarkers and risk-stratify patients from gigapixel-sized images. Machine learning problems in medical imaging often deal with rare diseases, making it important for these models to work in a label-imbalanced setting. Furthermore, these imbalances can occur in out-of-distribution (OOD) datasets when the models are deployed in the real-world. We leverage the idea that decoupling feature and classifier learning can lead to improved decision boundaries for label imbalanced datasets. To this end, we investigate the integration of supervised contrastive learning with multiple instance learning (SC-MIL). Specifically, we propose a joint-training MIL framework in the presence of label imbalance that progressively transitions from learning bag-level representations to optimal classifier learning. We perform experiments with different imbalance settings for two well-studied problems in cancer pathology: subtyping of non-small cell lung cancer and subtyping of renal cell carcinoma. SC-MIL provides large and consistent improvements over other techniques on both in-distribution (ID) and OOD held-out sets across multiple imbalanced settings.

Dinkar Juyal, Siddhant Shingi, Syed Ashar Javed, Harshith Padigela, Chintan Shah, Anand Sampat, Archit Khosla, John Abel, Amaro Taylor-Weiner

2023-03-23

General General

Deep learning for pancreatic diseases based on endoscopic ultrasound: A systematic review.

In International journal of medical informatics ; h5-index 49.0

BACKGROUND AND AIMS : Endoscopic ultrasonography (EUS) is one of the main examinations in pancreatic diseases. A series of the studies reported the application of deep learning (DL)-assisted EUS in the diagnosis of pancreatic diseases. This systematic review is to evaluate the role of DL algorithms in assisting EUS diagnosis of pancreatic diseases.

METHODS : Literature search were conducted in PubMed and Semantic Scholar databases. Studies that developed DL models for pancreatic diseases based on EUS were eligible for inclusion. This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines and quality assessment of the included studies was performed according to the IJMEDI checklist.

RESULTS : A total of 23 studies were enrolled into this systematic review, which could be categorized into three groups according to computer vision tasks: classification, detection and segmentation. Seventeen studies focused on the classification task, among which five studies developed simple neural network (NN) models while twelve studies constructed convolutional NN (CNN) models. Three studies were concerned the detection task and five studies were the segmentation task, all based on CNN architectures. All models presented in the studies performed well based on EUS images, videos or voice. According to the IJMEDI checklist, six studies were recognized as high-grade quality, with scores beyond 35 points.

CONCLUSIONS : DL algorithms show great potential in EUS images/videos/voice for pancreatic diseases. However, there is room for improvement such as sample sizes, multi-center cooperation, data preprocessing, model interpretability, and code sharing.

Yin Minyue, Liu Lu, Gao Jingwen, Lin Jiaxi, Qu Shuting, Xu Wei, Liu Xiaolin, Xu Chunfang, Zhu Jinzhou

2023-Mar-18

Convolutional neural networks, Deep learning, Endoscopic ultrasonography, Pancreatic diseases, Systematic review

Dermatology Dermatology

HGM-cNet: Integrating hippocampal gray matter probability map into a cascaded deep learning framework improves hippocampus segmentation.

In European journal of radiology ; h5-index 47.0

A robust cascaded deep learning framework with integrated hippocampal gray matter (HGM) probability map was developed to improve the hippocampus segmentation (called HGM-cNet) due to its significance in various neuropsychiatric disorders such as Alzheimer's disease (AD). Particularly, the HGM-cNet cascaded two identical convolutional neural networks (CNN), where each CNN was devised by incorporating Attention Block, Residual Block, and DropBlock into the typical encoder-decoder architecture. The two CNNs were skip-connected between encoder components at each scale. The adoption of the cascaded deep learning framework was to conveniently incorporate the HGM probability map with the feature map generated by the first CNN. Experiments on 135T1-weighted MRI scans and manual hippocampal labels from publicly available ADNI-HarP dataset demonstrated that the proposed HGM-cNet outperformed seven multi-atlas-based hippocampus segmentation methods and six deep learning methods under comparison in most evaluation metrics. The Dice (average > 0.89 for both left and right hippocampus) was increased by around or more than 1% over other methods. The HGM-cNet also achieved a superior hippocampus segmentation performance in each group of cognitive normal, mild cognitive impairment, and AD. The stability, conveniences and generalizability of the cascaded deep learning framework with integrated HGM probability map in improving hippocampus segmentation was validated by replacing the proposed CNN with 3D-UNet, Atten-UNet, HippoDeep, QuickNet, DeepHarp, and TransBTS models. The integration of the HGM probability map in the cascaded deep learning framework was also demonstrated to facilitate capturing hippocampal atrophy more accurately than alternative methods in AD analysis. The codes are publicly available at https://github.com/Liu1436510768/HGM-cNet.git.

Zheng Qiang, Liu Bin, Gao Yan, Bai Lijun, Cheng Yu, Li Honglun

2023-Mar-15

Deep learning, Gray matter volume, Hippocampus, Image segmentation, Multi-atlas segmentation

Public Health Public Health

Application of a developed triple-classification machine learning model for carcinogenic prediction of hazardous organic chemicals to the US, EU, and WHO based on Chinese database.

In Ecotoxicology and environmental safety ; h5-index 67.0

Cancer, the second largest human disease, has become a major public health problem. The prediction of chemicals' carcinogenicity before their synthesis is crucial. In this paper, seven machine learning algorithms (i.e., Random Forest (RF), Logistic Regression (LR), Support Vector Machines (SVM), Complement Naive Bayes (CNB), K-Nearest Neighbor (KNN), XGBoost, and Multilayer Perceptron (MLP)) were used to construct the carcinogenicity triple classification prediction (TCP) model (i.e., 1A, 1B, Category 2). A total of 1444 descriptors of 118 hazardous organic chemicals were calculated by Discovery Studio 2020, Sybyl X-2.0 and PaDEL-Descriptor software. The constructed carcinogenicity TCP model was evaluated through five model evaluation indicators (i.e., Accuracy, Precision, Recall, F1 Score and AUC). The model evaluation results show that Accuracy, Precision, Recall, F1 Score and AUC evaluation indicators meet requirements (greater than 0.6). The accuracy of RF, LR, XGBoost, and MLP models for predicting carcinogenicity of Category 2 is 91.67%, 79.17%, 100%, and 100%, respectively. In addition, the constructed machine learning model in this study has potential for error correction. Taking XGBoost model as an example, the predicted carcinogenicity level of 1,2,3-Trichloropropane (96-18-4) is Category 2, but the actual carcinogenicity level is 1B. But the difference between Category 2 and 1B is only 0.004, indicating that the XGBoost is one optimum model of the seven constructed machine learning models. Besides, results showed that functional groups like chlorine and benzene ring might influence the prediction of carcinogenic classification. Therefore, considering functional group characteristics of chemicals before constructing the carcinogenicity prediction model of organic chemicals is recommended. The predicted carcinogenicity of the organic chemicals using the optimum machine leaning model (i.e., XGBoost) was also evaluated and verified by the toxicokinetics. The RF and XGBoost TCP models constructed in this paper can be used for carcinogenicity detection before synthesizing new organic substances. It also provides technical support for the subsequent management of organic chemicals.

Hao Ning, Sun Peixuan, Zhao Wenjin, Li Xixi

2023-Mar-20

Carcinogenic chemicals, Carcinogenicity classification prediction model, Machine learning, Model evaluation metrics, Molecular structure, Toxicokinetics

General General

Development of prediction models for one-year brain tumour survival using machine learning: a comparison of accuracy and interpretability.

In Computer methods and programs in biomedicine

BACKGROUND AND OBJECTIVE : Prediction of survival in patients diagnosed with a brain tumour is challenging because of heterogeneous tumour behaviours and treatment response. Advances in machine learning have led to the development of clinical prognostic models, but due to the lack of model interpretability, integration into clinical practice is almost non-existent. In this retrospective study, we compare five classification models with varying degrees of interpretability for the prediction of brain tumour survival greater than one year following diagnosis.

METHODS : 1028 patients aged ≥16 years with a brain tumour diagnosis between April 2012 and April 2020 were included in our study. Three intrinsically interpretable 'glass box' classifiers (Bayesian Rule Lists [BRL], Explainable Boosting Machine [EBM], and Logistic Regression [LR]), and two 'black box' classifiers (Random Forest [RF] and Support Vector Machine [SVM]) were trained on electronic patients records for the prediction of one-year survival. All models were evaluated using balanced accuracy (BAC), F1-score, sensitivity, specificity, and receiver operating characteristics. Black box model interpretability and misclassified predictions were quantified using SHapley Additive exPlanations (SHAP) values and model feature importance was evaluated by clinical experts.

RESULTS : The RF model achieved the highest BAC of 78.9%, closely followed by SVM (77.7%), LR (77.5%) and EBM (77.1%). Across all models, age, diagnosis (tumour type), functional features, and first treatment were top contributors to the prediction of one year survival. We used EBM and SHAP to explain model misclassifications and investigated the role of feature interactions in prognosis.

CONCLUSION : Interpretable models are a natural choice for the domain of predictive medicine. Intrinsically interpretable models, such as EBMs, may provide an advantage over traditional clinical assessment of brain tumour prognosis by weighting potential risk factors and their interactions that may be unknown to clinicians. An agreement between model predictions and clinical knowledge is essential for establishing trust in the models decision making process, as well as trust that the model will make accurate predictions when applied to new data.

Charlton Colleen E, Poon Michael T C, Brennan Paul M, Fleuriot Jacques D

2023-Mar-13

Bayesian rule lists, Brain cancer, Explainable boosting machine, Interpretable models, Machine learning, Survival

General General

Forecasting and Optimizing Dual Media Filter Performance via Machine Learning.

In Water research

Four different machine learning algorithms, including Decision Tree (DT), Random Forest (RF), Multivariable Linear Regression (MLR), Support Vector Regressions (SVR), and Gaussian Process Regressions (GPR), were applied to predict the performance of a multi-media filter operating as a function of raw water quality and plant operating variables. The models were trained using data collected over a seven year period covering water quality and operating variables, including true colour, turbidity, plant flow, and chemical dose for chlorine, KMnO4, FeCl3, and Cationic Polymer (PolyDADMAC). The machine learning algorithms have shown that the best prediction is at a 1-day time lag between input variables and unit filter run volume (UFRV). Furthermore, the RF algorithm with grid search using the input metrics mentioned above with a 1-day time lag has provided the highest reliability in predicting UFRV with a RMSE and R2 of 31.58 and 0.98, respectively. Similarly, RF with grid search has shown the shortest training time, prediction accuracy, and forecasting events using a ROC-AUC curve analysis (AUC over 0.8) in extreme wet weather events. Therefore, Random Forest with grid search and a 1-day time lag is an effective and robust machine learning algorithm that can predict the filter performance to aid water treatment operators in their decision makings by providing real-time warning of the potential turbidity breakthrough from the filters.

Moradi Sina, Omar Amr, Zhou Zhuoyu, Agostino Anthony, Gandomkar Ziba, Bustamante Heriberto, Power Kaye, Henderson Rita, Leslie Greg

2023-Mar-12

Filtration performance, Hyper-parameter optimisation, Machine learning approach, Unit filter run volume