Core Technologies

Based on the foundation of proprietary technologies and a core “brain” built on a deep learning platform, SenseTime has rapidly opened up AI application in multiple vertical scenarios.

Medical Image Analysis

Medical Image Analysis

Technical Capabilities

Detection and Localization of Lesion/Body Part
Classification of Lesion
Differentiation of Benign and Malignancy & Grading of Disease
Lesion/Body Part Segmentation and Quantitative Analysis
Multi-modality Image Registration

Detection and Localization of Lesion/Body Part

Based on cutting-edge computer vision technology, our algorithm can swiftly detect and localize various lesions and organs in multi-modality medical images, such as detecting spine discs and vertebrae in CT or MRI, lung nodules in CT, cells in pathology images. It provides doctors with valuable references for diagnosis and therefore improve efficiency.

Classification of Lesion

By integrating international common practice of diagnosis guidelines and the experience of top medical professionals, our algorithm can precisely classify lesions, which plays a key role in clinical diagnoses and helps minimize misdiagnosis of similar lesions.

Differentiation of Benign and Malignancy & Grading of Disease

Based on the analysis of medical images and clinical factors, our algorithm can diagnose the malignancy and severity of various diseases, such as the grading of anterior cruciate ligament tear, the discrimination of benign and malignant lung nodules, and to meet the need of mass screening and hierarchical diagnosis.

Lesion/Body Part Segmentation and Quantitative Analysis

Our algorithm performs training based on small data sets to achieve pixel-level precise segmentation of multiple lesions and organs; conducts key parameter measurement and automatic, quantified analysis, such as radiotherapy target area delineation and pelvic tumor segmentation. This does not only release doctors from time-consuming, labor-intensive manual illustration work, but also fulfil the requirements of quantified diagnosis and personalized surgery planning scenarios.

Multi-modality Image Registration

With the registration of multi-modality data such as CT, MRI, and PET of the same body part or organ, our algorithm can achieve accurate fusion of different modalities and different sequence data, enabling more accurate qualitative grading and quantitative analysis of lesions.