Renato Antonio Krohling

To the best of our knowledge, existing benchmarks for bias mitigation in skin cancer classification focus exclusively on image-based, leaving a gap in the tabular-multimodal setting, where image-derived probabilities are combined with structured clinical data. To address this gap, this paper proposes an experimental framework to benchmark and integrate multiple bias mitigation algorithms into automated skin cancer classification pipelines based on a Mixed Fusion architecture. Using a multimodal fusion architecture originally developed for interpretability, we extend the framework to assess diagnostic bias across image data and metadata. Experiments were conducted on three datasets: PAD-UFES-20, HIBA, and MIDAS, incorporating sensitive demographic attributes to enable a fairness evaluation. Bias mitigation techniques were systematically applied at the pre-processing, in-processing, and post-processing stages, as well as in combination, using Multi-layer Perceptron (MLP), K-Nearest Neighbors (KNN), and Decision Tree (DT) classifiers. Performance and fairness were jointly analyzed using multiple metrics. According to the A-TOPSIS ranking, the MLP achieved the best results on DB-PAD-UFES-20, with a Balanced Accuracy (BACC) of 0.8313 and Statistical Parity (SP) of 0.0050 for gender and 0.0580 for Fitzpatrick skin type. On DB-HIBA, KNN ranked first, reaching a BACC of 0.7922 and SP of 0.0047 for gender. For DB-MIDAS, MLP again performed best, achieving a BACC of 0.6911 and SP of 0.0096 for gender. Post-processing methods consistently demonstrated effectiveness across datasets. By providing a methodological foundation with baseline results, this work paves the way for future research to develop more reliable clinical diagnosis that incorporate other bias mitigation techniques.

In this paper, we propose an approach to support the diagnosis of urinary tract diseases, with a focus on bladder cancer, using SHAP (SHapley Additive exPlanations)-based feature selection to enhance the transparency and effectiveness of predictive models. Six binary classification scenarios were developed to distinguish bladder cancer from other urological and oncological conditions. The algorithms XGBoost, LightGBM, and CatBoost were employed, with hyperparameter optimization performed using Optuna and class balancing with the SMOTE technique. The selection of predictive variables was guided by importance values through SHAP-based feature selection while maintaining or even improving performance metrics such as balanced accuracy, precision, and specificity. The use of explainability techniques (SHAP) for feature selection proved to be an effective approach. The proposed methodology may contribute to the development of more transparent, reliable, and efficient clinical decision support systems, optimizing screening and early diagnosis of urinary tract diseases.

The automated diagnosis of skin cancer is a very challenging task. Aiming to imitate dermatologists, computer aided diagnostic (CAD) of skin cancer based on artificial intelligence (AI) provide performance close or superior to dermatologist with clinical images acquired from smartphones taking into account also the patient lesion information. Although the addition of patient's lesion information as if the lesion bleeds, or hurts, or if it itches, among others, do improve the diagnosis metrics, they are not enough to correctly discriminate the main skin cancer lesions. The gold-standard in clinical diagnosis is the biopsy exam. For that, data was collected from 2018 in partnership with the university hospital at UFES, resulting in more than 2000 clinical images of the main skin lesions taken by smarpthones with the corresponding patient lesion information. The main value of the dataset is that it is certified with the histopathological exams (gold-standard). The dataset PAD-UFES-20 is one of the real-world dataset used in this area since 2020. Now, we propose a new methodology to use the information of the histopathological images to automated diagnosis. Our approach to solve the problem uses a convolutional neural network (CNN) to extract features of the clinical images and a second one to extract features of the corresponding histopathological images and a feedforward neural network for feature mapping, which learns to map the extracted clinical features in its corresponding histopathological features. Next, the entire framework for training the model firstly extracts features of the clinical images and use the previously trained neural network for feature mapping to provide the corresponding histopathological features. Both features (extracted and mapped) are fused and inputted to a neural network classifier providing the final diagnosis.

Skin lesions are classified in benign or malignant. Among the malignant, melanoma is a very aggressive cancer and the major cause of deaths. So, early diagnosis of skin cancer is very desired. In the last few years, there is a growing interest in computer aided diagnostic (CAD) using most image and clinical data of the lesion. These sources of information present limitations due to their inability to provide information of the molecular structure of the lesion. NIR spectroscopy may provide an alternative source of information to automated CAD of skin lesions. The most commonly used techniques and classification algorithms used in spectroscopy are Principal Component Analysis (PCA), Partial Least Squares - Discriminant Analysis (PLS-DA), and Support Vector Machines (SVM). Nonetheless, there is a growing interest in applying the modern techniques of machine and deep learning (MDL) to spectroscopy. One of the main limitations to apply MDL to spectroscopy is the lack of public datasets. Since there is no public dataset of NIR spectral data to skin lesions, as far as we know, an effort has been made and a new dataset named NIR-SC-UFES, has been collected, annotated and analyzed generating the gold-standard for classification of NIR spectral data to skin cancer. Next, the machine learning algorithms XGBoost, CatBoost, LightGBM, 1D-convolutional neural network (1D-CNN) were investigated to classify cancer and non-cancer skin lesions. Experimental results indicate the best performance obtained by LightGBM with pre-processing using standard normal variate (SNV), feature extraction providing values of 0.839 for balanced accuracy, 0.851 for recall, 0.852 for precision, and 0.850 for F-score. The obtained results indicate the first steps in CAD of skin lesions aiming the automated triage of patients with skin lesions in vivo using NIR spectral data.