Time to read
2 minutes
Read so far

AI technique can detect colorectal cancerous tissue in real time

Wed, 06/02/2021 - 09:17
Posted in:

A surgical technique developed at University College Dublin uses artificial intelligence to detect cancerous tissue in real time during surgery could radically improve health outcomes. The new method demonstrates how with the use of a digital camera and dyes, cancer processes in living tissue can be viewed during an operation, allowing surgeons to see the exact extent of cancers during a procedure, ensuring that the maximum amount of cancerous tissue is surgically removed. In the study, ‘Digital dynamic discrimination of primary colorectal cancer using systemic indocyanine green with near-infrared endoscopy’, published in Nature Scientific Reports.

Previously surgeons had considerable wait times before a formal characterisation of tissue types could be performed by laboratories. This delay also occurred when assessing responsiveness to medical therapies by interval radiological imaging.

"If cancer can be fully detected, it's much more likely to be cured in one single operation or have combination therapies better sequenced and so the risk to the patient of recurrence and complications are markedly reduced," said Ronan Cahill, Professor of Surgery at the UCD School of Medicine and the Mater Misericordiae University Hospital (MMUH). "Dynamic digital discrimination of cancer right at the time of intervention means the surgical team can better perfect the right intervention to the individual patient first time."

Indocyanine green (ICG) with near-infrared (NIR) endoscopy is commonly used to enhance real-time intraoperative tissue microperfusion appreciation. However, Cahill and colleagues hypothesised that it may also dynamically reveal neoplasia distinctively from normal tissue especially with video software fluorescence analysis. In the study, colorectal tumours of patients were imaged mucosally following ICG administration (0.25mg/kg IV) using an endo-laparoscopic NIR system (PINPOINT Endoscopic Fluorescence System, Stryker) including immediate, continuous in situ visualisation of rectal lesions transanally for up to 20 minutes.

Both spot and dynamic temporal fluorescence intensities (FI) were quantified using ImageJ (including videos at one frame/second, fps) and by a bespoke MATLAB application that provided digitalised video tracking and signal logging at 30fps (Fluorescence Tracker App downloadable via MATLAB file exchange). They them performed statistical analyses of FI-time plots comparing tumours (benign and malignant) against control during FI curve rise, peak and decline from apex. Early kinetic FI signal measurement delineated discriminative temporal signatures from tumours (n=20, nine cancers) offering rich data for analysis versus delayed spot measurement (n=10 cancers).

The outcomes showed that malignant lesion dynamic curves peaked significantly later with a shallower gradient than normal tissue while benign lesions showed significantly greater and faster intensity drop from apex versus cancer. The researchers explained that an analysis of a continuous stream of intraoperatively acquired early ICG fluorescence data can act as an in situ tumour-identifier with greater detail than later snapshot observation alone. In addition, software quantification of such kinetic signatures may distinguish invasive from non-invasive neoplasia with potential for real-time in silico diagnosis.

"The tools we are developing are straightforward to deploy and use software to allow users easily interpret the findings without having to develop further specialist knowledge," added Cahill.  "A few minutes is enough to determine if a lesion is cancerous. If it is there, there is no need to wait for a biopsy, we can remove it straight away. We also have a better chance of getting all of the cancer out first time and increasing a person's chances of a cure," Cahill told the Irish Times (https://www.irishtimes.com/news/health/new-tumour-removal-surgery-using…)

The scientific work on the new technique was done at UCD and the Mater Hospital, with technological collaboration from IBM Research, and method has been successfully applied to 200 patients.

The new approach is particularly effective for colorectal cancers, said Mr Jeffrey Dalli, General Surgeon and Surgical Research Fellow at UCD, who added that method detects cancerous tissue not just by its appearance but by its behaviour - allowing it to be clearly distinguished from nearby normal tissue.

"Colon and rectal cancers are common, being the second most common major cancer type in men and women and they are increasing in incidence especially among younger people," he said. "This technology will help surgeons better discriminate during operations exactly what is best for each individual patient."

To access this paper, please here