Skin cancer: Brazilian AI speeds up lesion screening using cell phone photos

A new artificial intelligence tool, developed by national startup AI Pathology, promises to optimize the skin lesion screening process using only photos taken with a cell phone, which contributes to the early identification of skin cancer. The innovation called Nevo, operates with technological support from Nvidia.
In collaboration with the National Rural Learning Service (Senar) of Goiás, the technology was applied in a project that resulted in the detection of 75 occurrences, later validated by medical diagnoses. For seven months, the tool helped to evaluate 2,078 rural workers in the region.
How Nevo artificial intelligence helps with detection
Nevo does not require complex devices to operate, analyzing images captured by smartphones to look for skin lesions that indicate risks. This technology speeds up tracking and organizes the priority of clinical care according to the malignancy potential of each case.
- The platform combines photos of injuries with important data, such as the affected area, the individual’s health history and personal information;
- Next, AI provides a quick and safe analysis, according to Brazilian healtech, which allows you to quickly identify situations that require attention or a biopsy;
- It is essential to understand that the tool does not issue diagnoses, this being the exclusive responsibility of the doctor. The system classifies the risk of the injury and suggests referral to a dermatologist;
- This method reduced screening time by 90% when compared to traditional processes, managing to analyze more than 400 images per hour, which makes it ideal for places with high demand for services.
With the ability to examine an image in less than 8 seconds, Nevo AI operates in an environment optimized by graphics processing units (GPUs). This system offers the computational power necessary to train and apply high-quality medical deep learning models, with convolutional neural networks (RNCs) as its technological base.
The platform’s algorithms, which are inspired by the structure of the human visual cortex, can distinguish morphological, texture and structure changes, in addition to the color distribution of the lesions. In this way, they identify characteristics that indicate whether the lesions are benign or malignant, achieving an accuracy greater than 94%, according to the company.

Free treatment for patients benefiting from screening
As part of the collaboration with Senar de Goiás, people who had their lesions classified by AI as possible cases of cancer were referred to specialists at the Skin Cancer Institute of Goiânia. There, they underwent detailed examinations and received free treatment.
The company AI Pathology has plans to expand the use of Nevo to other regions of the country, especially where there is a higher incidence of skin cancer and strong sun exposure. The long-term vision includes creating a worldwide network of artificial intelligence-based systems to drive preventative medicine and early diagnosis.






