A technological revolution in early colon cancer diagnosis through 3D models
Recognising problem areas and supporting doctors in the investigation of suspected colon cancer: this is the aim of the project Real SynCol: a high fidelity synthetic colon dataset for 3D reconstruction applications, presented during the Third General Meeting of Musa. ‘Our challenge is to assist the doctor in the diagnosis of colon cancer, where one of the main problems is that part of the organ’s mucosa is not visualised during the examination,’ explains Chiara Lena, PhD student at the Politecnico di Milano and head of the Real SynCol project.
To support doctors at this critical stage, Real SynCol aims to develop a 3D model of the colon during the examination, identifying areas where ‘holes’ appear in the reconstruction, suggesting potential hidden lesions. The artificial intelligence algorithms required to obtain a complete model require a large amount of data. ‘What we did was to create a virtual colon model in which we could reconstruct all this data to support the first phase of artificial intelligence algorithm development,’ Lena continues.
The impact of this study is significant, as very few algorithms are currently able to provide a reliable reconstruction of the colon. Using a virtual dataset, this additional data can ‘teach’ the algorithms to improve reconstruction of missing portions and estimate how much of the colon has been visualised, improving the quality of the analysis. ‘It is estimated that almost a quarter of the colon is not visible during the examination; being able to reduce this portion could have a huge impact. Colon cancer develops slowly, so detecting lesions early is crucial,’ says Lena. ‘Early and effective screening significantly reduces the negative impact of neglected lesions on the patient. This is the goal of our study.’
