Breast cancer is the most common malignancy in women worldwide, and is responsible for more than half a million deaths each year. The appropriate therapy depends on the evaluation of the expression of various biomarkers, such as the human epidermal growth factor receptor 2 (HER2) transmembrane protein, through specialized techniques, such as immunohistochemistry or in situ hybridization. In this work, we present the HER2 on hematoxylin and eosin (HEROHE) challenge, a parallel event of the 16th European Congress on Digital Pathology, which aimed to predict the HER2 status in breast cancer based only on hematoxylin–eosin-stained tissue samples, thus avoiding specialized techniques. The challenge consisted of a large, annotated, whole-slide images dataset (509), specifically collected for the challenge. Models for predicting HER2 status were presented by 21 teams worldwide. The best-performing models are presented by detailing the network architectures and key parameters. Methods are compared and approaches, core methodologies, and software choices contrasted. Different evaluation metrics are discussed, as well as the performance of the presented models for each of these metrics. Potential differences in ranking that would result from different choices of evaluation metrics highlight the need for careful consideration at the time of their selection, as the results show that some metrics may misrepresent the true potential of a model to solve the problem for which it was developed. The HEROHE dataset remains publicly available to promote advances in the field of computational pathology.

Objectives. Evidence suggests that about 80 of all salivary gland tumors involve the parotid glands, with approximately 20 of parotid gland tumors (PGTs) being malignant. Discriminating benign and malignant parotid gland lesions preoperatively is vital for selecting the appropriate treatment strategy. This study explored the diagnostic performance of deep learning system for discriminating benign and malignant PGTs in ultrasonography images and compared it with radiologists. Methods. A total of 251 consecutive patients with surgical resection and proven parotid gland malignant or benign tumors who underwent preoperative ultrasound examinations were enrolled in this study between January 2014 and November 2020. Next, we compared the diagnostic accuracy of deep learning methods (ViT-B\16, EfficientNetB3, DenseNet121, and ResNet50) and radiologists in parotid gland tumor. In addition, the area under the curve (AUC), specificity, sensitivity, positive predictive value, and negative predictive value were calculated. Results. Among the 251 patients, 176/251 were the training set, whereas 75/251 were the validation set. Results showed that 74/251 patients had malignant tumor. Deep learning models achieved good performance in differentiating benign from malignant tumors, with the diagnostic accuracy and AUCs of ViT-B\16, EfficientNetB3, DenseNet121, and ResNet50 model being 81 and 0.81, 80 and 0.82, 77 and 0.81, and 79 and 0.80, respectively. On the other hand, the diagnostic accuracy and AUCs of radiologists were 77-81 and 0.68-0.75, respectively. It was evident that the diagnostic accuracy of deep learning methods was higher than that of inexperienced radiologists, but there was no significant difference between deep learning methods and experienced radiologists. Conclusions. This study shows that the deep learning system can be used for diagnosing parotid tumors. The findings also suggest that the deep learning system may improve the diagnosis performance of inexperienced radiologists.

Organic optoelectronic materials (OOMs) are pivotal for advancing technologies such as organic photovoltaics and light-emitting diodes. Traditional methods for discovering new OOMs are inefficient and limited by chemical space exploration. We introduce O2-GEN, a novel framework leveraging a 3D pretraining backbone trained on a diverse dataset of over ten million molecules, enabling comprehensive exploration of chemical space. O2-GEN is effective at generating novel fused-ring systems and conjugated fragment assemblies. It significantly outperforms existing models in speed and chemical structural validity, particularly for larger molecules. The framework supports both global and local generation modes, allowing for the design of new molecules or modifications of existing molecules. Additionally, O2-GEN integrates a property selector fine-tuned with density functional theory data, enabling precise multi-property screening. Overall, our O2-GEN allows the construction of datasets with multi-property-biased distributions tailored to specific application scenarios, facilitating the discovery of novel optoelectronic materials.