a) A zoomed-out view of a papillary thyroid carcinoma tissue sample in which...
a) A zoomed-out view of a papillary thyroid carcinoma tissue sample in which the results of the deep learning (DL) algorithm’s first layer are shown (blue). From the registered carcinoma area, the DL algorithm then registers the carcinoma epithelium as either b) tall cell (TC) (red) or c) non-TC area (green). The DL algorithm determines the percentage of the epithelium covered by TCs, that is, the TC score.

Image source: Stenman et al., PLOS One 2022 (CC BY 4.0)

Supervised learning approach

A new deep learning-based algorithm to predict relapse-free survival in papillary thyroid carcinoma

According to the World Health Organization, the tall cell variant (TCV) is an aggressive subtype of papillary thyroid carcinoma (PTC) featuring at least 30% epithelial cells two to three times longer than they are wide. In practice, applying this distinction is difficult, leading to substantial variations between observers.

‘That’s why we are developing and training a deep learning algorithm using supervised learning to detect and quantify the proportion of tall cells (TCs) in papillary thyroid carcinoma,’ explained Sebastian Stenman, researcher from the Institute for Molecular Medicine, and the Department of Pathology at the University of Helsinki, Finland. 

We showed that the DL-based algorithm was better than the human observer in identifying tall cell variants

Sebastian Stenman

In summer 2022, his research team is testing it on an independent data set, and further validating it on an independent set of 90 papillary thyroid carcinoma samples from patients treated at the Hospital District of Helsinki and Uusimaa (HUS) between 2003 and 2013. The Finnish scientist compared the algorithm-based tall cell ratio to independent scoring by a human investigator and looked at how those scores were associated with disease outcomes. 

The results published revealed that the deep learning algorithm detected tall cells with a sensitivity of 93.7% and a specificity of 94.5%.  In the validation set, the deep learning algorithm tall cell scores correlated with a diminished relapse-free survival.  ‘We showed that the DL-based algorithm was better than the human observer in identifying tall cell variants,’ said Stenman. The algorithm could prove useful as a clinical tool for pathologists when evaluating PTC samples and could potentially significantly improve the consistency of TCV case assessment. So far, no such algorithm has been described. 


Read all latest stories

Related articles


Article • AI provides prognostic information

Next-generation deep learning models predict cancer survival

Deaths from cancer are currently estimated at 10 million each year worldwide. Conventional cancer staging systems aim to categorize patients into different groups with distinct outcomes. ‘However,…


News • Primary tumor or metastasis?

Deep learning and radiomics for precise differentiation in brain tumors

The distinction between primary tumors and metastases can be made quickly and accurately in brain tumors using radiomics and deep learning algorithms, a new study shows.


News • Application in pathology

Endometrial cancer diagnostics: AI gives new insights

New research shows the power of artificial intelligence (AI) applied to endometrial carcinoma microscopy images. This could improve diagnosis and treatment of uterine cancer.

Related products

Advanced intelligent Clear-IQ Engine for MR

Artificial Intelligence

Canon · Advanced intelligent Clear-IQ Engine for MR

Canon Medical Systems Europe B.V.
CT Image Reconstruction

Artificial Intelligence

Canon · CT Image Reconstruction

Canon Medical Systems Europe B.V.


Fujifilm · Dynamyx

HIT Automation Platform

Artificial Intelligence

Canon · HIT Automation Platform

Canon Medical Systems Europe B.V.

Artificial Intelligence

Fujifilm · REiLI

Vantage Elan NX Edition

1.5 Tesla

Canon · Vantage Elan NX Edition

Canon Medical Systems Europe B.V.
Subscribe to Newsletter