Background/Objectives: Identifying tumor budding (TB) in colorectal cancer (CRC) is vital for better prognostic assessment as it may signify the initial stage of metastasis. Despite its importance, TB detection remains challenging due to subjectivity in manual evaluations. Identifying TBs remains difficult, especially at high magnification levels, leading to inconsistencies in prognosis. To address these issues, we propose an automated system for TB classification using deep learning. Methods: We trained a deep learning model to identify TBs through weakly supervised learning by aggregating positive and negative bags from the tumor invasive front. We assessed various foundation models for feature extraction and compared their performance. Attention heatmaps generated by attention-based multi-instance learning (ABMIL) were analyzed to verify alignment with TBs, providing insights into the interpretability of the features. The dataset includes 29 WSIs for training and 70 whole slide images (WSIs) for the hold-out test set. Results: In six-fold cross-validation, Phikon-v2 achieved the highest average AUC (0.984 ± 0.003), precision (0.876 ± 0.004), and recall (0.947 ± 0.009). Phikon-v2 again achieved the highest AUC (0.979) and precision (0.980) on the external hold-out test set. Moreover, its recall rate (0.910) was still higher than that of UNI’s (0.879). UNI exhibited a balanced performance on the hold-out test set, with an AUC of 0.960 and a precision of 0.968. CtransPath showed strong precision on the external hold-out test set (0.947) but had a slightly lower recall (0.911). Conclusions: The proposed technique enhances the accuracy of TB assessment, offering potential applications for CRC and other cancer types.

Dementia is an age-related neurological disease and gives rise to profound cognitive decline in patients’ life. Alzheimer’s Disease (AD) is the progression of dementia and AD patients generally have memory loss and behavioral disorders. It is possible to determine the stage of dementia by developing automated systems via. signals obtained from patients. EEG is a popular brain monitoring system due to its cost effective, non-invasive implementation, and higher time resolution. In current study, we include participants of 24 HC (12 eyes open (EO), 12 eyes closed (EC)), and 24 AD (HC (12 eyes open (EO), 12 eyes closed (EC)). The aim of current study is to design a practical AD detection tool for AD/HC participants with a model called DWT-CNN. We performed Discrete Wavelet Transform (DWT) to extract EEG sub-bands. A Conv2D architecture is applied to raw samples of related EEG sub-bands. According to obtained performance metrics calculated from confusion matrices, all AD and HC time series are correctly classified for alpha band and full band range under both EO and EC. Classification rate of AD vs. HC increases under EO state in all cases even if EC is commonly preferred in other studies. We will add MCI patients with equal size and similar demographics and repeat the experimental steps to develop early alert system in future studies. Adding more participants will also increase generalization ability of method. It is also promising study to combine EEG with different modalities (2D TF image conversion, or MRI) in a multimodal approach.

Background: Axillary lymph nodes (ALNs) are the most important prognostic factor for survival in breast cancer. Pathological evaluation can affect the number of involved lymph nodes. In the current study, we evaluated whether the metastatic lymph node ratio (n ratio) is important in predicting disease-free survival (DFS) for breast cancer patients. Material and Methods: From 802 breast cancer cases, 427 patients with ALN metastasis were analyzed retrospectively. The n ratio was categorized as n ratio 1 (1–10%), n ratio 2 (10.01–50%) and n ratio 3 (>50%). DFS was established according to the Kaplan-Meier method. Predicting risk factors for relapse were analyzed using the Cox proportional hazards model. Results: The n ratio was significantly higher in breast cancer patients with advanced pathologic pT, pN and clinical stage, undifferentiated histology, lymphovascular and extracapsular invasion, more resected ALNs and positive progesterone receptor. In the univariate analysis, multicentricity, necrosis, grade, pN stage, estrogen receptor and progesterone receptor positivity, trastuzumab and neoadjuvant chemotherapy usage, the presence of inflammatory breast cancer and n ratio were found to be important factors in predicting DFS. Multivariate analysis indicated that multicentricity, neoadjuvant chemotherapy, trastuzumab usage and n ratio were significantly associated with prognosis. Conclusions: The n ratio is inexpensive, easily available and a simple prognostic factor for breast cancer patients with positive ALNs.