Artificial intelligence–based technology for semi-automated segmentation of rectal cancer using high-resolution MRI

Abstract

Aim Although MRI has a substantial role in directing treatment decisions for locally advanced rectal cancer, precise interpretation of the findings is not necessarily available at every institution. In this study, we aimed to develop artificial intelligence-based software for the segmentation of rectal cancer that can be used for staging to optimize treatment strategy and for preoperative surgical simulation. Method Images from a total of 201 patients who underwent preoperative MRI were analyzed for training data. The resected specimen was processed in a circular shape in 103 cases. Using these datasets, ground-truth labels were prepared by annotating MR images with ground-truth segmentation labels of tumor area based on pathologically confirmed lesions. In addition, the areas of rectum and mesorectum were also labeled. An automatic segmentation algorithm was developed using a U-net deep neural network. Results The developed algorithm could estimate the area of the tumor, rectum, and mesorectum. The Dice similarity coefficients between manual and automatic segmentation were 0.727, 0.930, and 0.917 for tumor, rectum, and mesorectum, respectively. The T2/T3 diagnostic sensitivity, specificity, and overall accuracy were 0.773, 0.768, and 0.771, respectively. Conclusion This algorithm can provide objective analysis of MR images at any institution, and aid risk stratification in rectal cancer and the tailoring of individual treatments. Moreover, it can be used for surgical simulations.

The role of artificial intelligence in surgical simulation

“…As seen in Figure 1 , segmentation and anatomical labelling seems to be the major application of AI in surgical simulation; however, generalisability of these algorithms remains a challenge ( 32 ). Although segmentation has proven its efficacy in small clinical studies, its accuracy is yet to reach perfection ( 33 ).…”

“…In addition to overcoming these fundamental obstacles, a leap to translation will require succeeding platforms to be able to recognise and capture different frames from an entire simulation, in order to account for users' variety of approaches and techniques that can adequately fulfil a single metric (3). As seen in Figure 1, segmentation and anatomical labelling seems to be the major application of AI in surgical simulation; however, generalisability of these algorithms remains a challenge (32). Although segmentation has proven its efficacy in small clinical studies, its accuracy is yet to reach perfection (33).…”

“…Although segmentation has proven its efficacy in small clinical studies, its accuracy is yet to reach perfection (33). Studies reported that their algorithms failed to identify certain structures such as complex vasculature or certain types of tumours and that their accuracy can be highly dependent on the fidelity of different imaging modalities (32)(33)(34). Furthermore, to train AI algorithms to an absolute precision without bias and to be generalisable, experts are needed to construct comprehensive datasets which require numerous annotations and segmentation of anatomical structures.…”

Essential updates 2020/2021: Advancing precision medicine for comprehensive rectal cancer treatment

“…Artificial intelligence‐based technology has been constructed to visualize the rectal cancer area, which can potentially be utilized to find the malignant features of rectal cancer. 167 , 168 , 169 In addition, according to a recent report, artificial intelligence can be utilized for preoperative simulation to visualize anatomical configuration individually, possibly improving operative safety. 170 If the diagnostic accuracy could be more sophisticated, the treatment strategy may be optimized based on the status of the individual tumor, which would be beneficial for the patient.…”

“…In recent years, some attempts have been made to diagnose on MRI precisely. Artificial intelligence‐based technology has been constructed to visualize the rectal cancer area, which can potentially be utilized to find the malignant features of rectal cancer 167–169 . In addition, according to a recent report, artificial intelligence can be utilized for preoperative simulation to visualize anatomical configuration individually, possibly improving operative safety 170 .…”