Machine learning algorithms are increasingly being used in the medical field to detect cancer from histopathological images, as manual detection of cancer by expert pathologists and clinicians is time-consuming. The prognosis is worse when cancer has spread to the lymph nodes. Fortunately, with the rise of digital pathology, high-resolution digital images of histopathological scans of lymph nodes are increasingly available. Manual assessments of hematoxylin and eosin (H&E) suffer from indistinct guidable observability. To address this problem, whole slide images (WSIs) are obtained from glass histopathology with the aid of a scanning device that analyses the sample. The extraction of the nucleus enhances the diagnostic potential of computational pathology, as the WSI contains a significant number of clustered nuclei in which tumors are present, i.e., correctly segmenting the nucleus present in the cluster will allow for downstream analysis of tumor infiltration lymphocytes (TILS). To achieve roto-translation invariance in the features extracted from the histopathological images, group equivariant convolutions were employed. The utilization of deep learning models incorporating G-CNN layers significantly enhanced the ability to detect metastatic features. A final AUC score of 0.98 is achieved. In addition, to identify nucleus instances, an encoder-decoder architecture employing G-CNN layers has also been employed, resulting in a Dice coefficient of 0.64 and a PQ value of 0.25.