Colon cancer: AI-based marker can improve quality of treatment

A Norwegian-led research group has developed a clinically useful prognostic marker using deep learning and digital scanning of conventional hematoxylin and eosin-stained tumor tissue sections. The assay has been extensively evaluated in large, independent patient populations, correlates with and outperforms established molecular and morphological prognostic markers, and gives consistent results across tumor and nodal stage.

Photo
Usind deep learning and digital scanning of conventional hematoxylin and eosin-stained tumor tissue sections, researchers have developed a clinically useful prognostic marker .
Source: Institute for Cancer Genetics and Informatics, Oslo University Hospital

The biomarker stratified stage II and III patients into sufficiently distinct prognostic groups. The tool has the potential to guide the selection of adjuvant treatment by avoiding therapy in very low-risk groups, and at the same time, identifying patients who would benefit from more intensive treatment regimes. The new method, which detects the DoMore-v1-CRC marker, is called histotyping.

More than 12,000,000 image tiles from patients with a distinctly good or poor disease outcome from four cohorts were used to train a total of 10 convolutional neural networks purpose-built for classifying super-sized heterogeneous images. A prognostic biomarker integrating the 10 networks was determined using patients with a non-distinct outcome.

The researchers tested 920 patients with slides prepared in the UK, which were independently validated according to a predefined protocol in 1122 patients treated with single-agent capecitabine using slides prepared in Norway. All cohorts included only patients with resectable tumors, and a formalin-fixed, paraffin-embedded tumor tissue block available for analysis. The primary outcome was cancer-specific survival. Eight-hundred twenty-eight patients from four cohorts had a distinct outcome and were used as a training cohort to obtain clear ground truth. 1645 patients had a non-distinct outcome and were used for tuning.

The biomarker provided a hazard ratio for poor versus good prognosis of 3·84 (95% CI 2·72–5·43; p<0·0001) in the primary analysis of the validation cohort. After adjusting for established prognostic markers significant in univariable analyses of the same cohort, (pN stage, pT stage, lymphatic invasion, venous vascular invasion) was the hazard ratio 3·04 (2·07–4·47; p<0·0001).

The method can differentiate between different stages of cancer and between patients in equal stages, but with different prognoses, using deep learning and neural networks. With today's pathological methods, as many as 80% of patients end up with an unclear prognosis. With the new AI-based method, is this figure has reduced to 12%. The researchers behind the marker expect it to be a useful tool for the choice of treatment. "The prognostic tools we have today, especially in the most common cancers, are far too coarse. We can with more precise prognostication, avoid both over- and undertreatment. Both will be important for the patient and the clinical services," says Sigbjørn Smeland, director of the Cancer Clinic, Oslo University Hospital.

Subscribe to our newsletter

Related articles

Artificial intelligence shortcuts introduce bias in cancer treatment

Artificial intelligence shortcuts introduce bias in cancer treatment

AI tools models are a powerful tool in cancer treatment. However, unless these algorithms are properly calibrated, they can sometimes make inaccurate or biased predictions.

Artificial intelligence rapidly computes protein structures

Artificial intelligence rapidly computes protein structures

Scientists have created a deep learning method, RoseTTAFold, to provide access to highly accurate protein structure prediction.

MasSpec Pen shows promise in pancreatic cancer surgery

MasSpec Pen shows promise in pancreatic cancer surgery

The MasSpec Pen has shown to accurately identify tissues and surgical margins directly in patients and differentiate healthy and cancerous tissue from banked pancreas samples.

Deep learning Assesses risk of cardiovascular disease

Deep learning Assesses risk of cardiovascular disease

A team of engineers from Rensselaer Polytechnic Institute and clinicians from Massachusetts General Hospital developed a deep learning algorithm that can help assess a patient's risk of cardiovascular disease with the same low-dose computerized tomography (CT) scan used to screen for lung cancer.

Using polarized LEDs for detection of cancer progression

Using polarized LEDs for detection of cancer progression

Researchers have experimentally demonstrated a novel cancer diagnosis technique based on the scattering of circularly polarized light.

Radiation therapy: 3D printed shields for protection

Radiation therapy: 3D printed shields for protection

To reduce tissue injury side effects from radiation therapy, researchers have developed 3D-printed gastrointestinal radioprotective devices that can be generated from patient CT scans.

AI method can detect precursors to cervical cancer

AI method can detect precursors to cervical cancer

Using AI and mobile digital microscopy, researchers hope to create screening tools that can detect precursors to cervical cancer in women in resource-limited settings.

AI can identify cancerous cells by their acidity

AI can identify cancerous cells by their acidity

Using a special dye, cells are colored according to their pH, and a machine learning algorithm can detect changes in the color spectrum due to cancer.

AI, holographic microscopy beat scientists at analyzing immunotherapy​

AI, holographic microscopy beat scientists at analyzing immunotherapy​

AI is helping researchers decipher images from a new holographic microscopy technique needed to investigate a key process in cancer immunotherapy “live” as it takes place.

Popular articles

Subscribe to Newsletter