Thursday, March 28, 2024
HomeChildren's HealthAI-based danger mannequin for breast most cancers screening

AI-based danger mannequin for breast most cancers screening

A latest Lancet Regional Well being examine assesses the efficiency of a synthetic intelligence (AI)-based danger mannequin for breast most cancers screening in Europe.

Examine: European validation of an image-derived AI-based short-term danger mannequin for individualized breast most cancers screening—a nested case-control examine. Picture Credit score: Gagliardiphotography /


Common mammography screening has decreased deaths resulting from breast most cancers in ladies. Even after biennial screening for breast most cancers, about 25% of breast cancers are identified. In these circumstances, some ladies may need examined destructive in a single mammographic screening however might have been identified with breast most cancers earlier than attending their subsequent screening appointment.

Between 25-40% of girls are identified with breast most cancers at stage two or increased. Thus, you will need to decide whether or not the tumor was detected through the common mammographic screening, as it’s a strong prognostic marker of breast cancer-related mortality.

Earlier research have proposed the addition of different danger evaluation measures to enhance the screening course of and in the end forestall the chance of interval most cancers earlier than the subsequent display screen. This technique might additionally cut back the incidence of late-stage breast most cancers within the subsequent display screen. In the USA, ladies who’ve dense breasts or are at a excessive danger resulting from familial danger components, bear extra examinations.

The present breast most cancers screening applications performed in Europe shouldn’t have any tips that point out the efficiency of extra examinations for girls at the next danger of breast most cancers. Nevertheless, a number of medical danger evaluation instruments have been developed based mostly on household historical past and way of life components to enhance screening outcomes.

Though a brand new image-based danger mannequin has proven appreciable potential in figuring out ladies at the next danger of breast most cancers, this mannequin requires extra exterior validation to evaluate its medical feasibility.

Concerning the examine

The present examine assessed a beforehand developed image-derived AI-based danger mannequin for breast most cancers that was designed to establish the chance of breast most cancers within the quick time period. Extra particularly, this mannequin has been used to establish ladies who developed most cancers within the interval between two mammography screenings in two years after a destructive display screen.

The general danger classification and discriminatory efficiency of the ProFound AI Threat mannequin have been assessed. This AI-based mannequin was beforehand developed utilizing a screening Swedish cohort.

The present examine used 4 screening populations comprising ladies between 45 and 69 years of age who underwent mammographic screening. From this screening inhabitants, two cohorts have been designed in Germany and one every from Italy and Spain.

Among the key eligibility standards included the incidence of breast most cancers with a digital mammogram at baseline. These ladies have been identified earlier than or on the subsequent screening program. 

The examine excluded ladies with a household historical past of breast most cancers. A nested case-control examine for every inhabitants was carried out. Management teams for every screening inhabitants have been randomly designed from the underlying screening cohort.

Examine findings

The validation examine included a complete of 739 breast most cancers sufferers and seven,812 controls. The most cancers final result was assessed on the second display screen, throughout which ladies have been randomly assigned to have digital mammography or have been subjected to digital breast tomosynthesis (DBT). The AI-based danger mannequin used these mammographs to foretell ladies who have been prone to breast most cancers in two years.

As in comparison with the unique evaluation of the AI-based danger mannequin for breast most cancers screening that used a Swedish cohort, a small variability of discriminatory performances throughout populations of various European international locations was noticed. Nevertheless, the mannequin exhibited related discrimination to that of the earlier report. Girls with dense and non-dense breasts exhibited related danger stratification efficiency.

Superior-stage breast most cancers was almost certainly to be identified in high-risk ladies as in comparison with these at a reasonable danger of growing breast most cancers. The present examine indicated that an image-based AI-risk mannequin may very well be affected by ethnic variations and screening frequencies.

Girls with non-dense breasts have been discovered to be at a larger danger of growing extra aggressive interval cancers. In distinction, ladies with dense breasts might have their tumor masked by dense tissue, which will increase the potential of growing interval most cancers and late-stage breast most cancers.

Radiologists expertise vital challenges associated to the masking of tumors by dense tissues. Due to this fact, high-risk ladies with dense breasts might positively profit from extra delicate examinations following a destructive screening. Nonetheless, a shorter screening interval is preferable for high-risk ladies with non-dense breasts because of the elevated danger of a fast-growing tumor. 


The present examine supplied insights into the significance of conducting extra checks past mammographic density to establish ladies who’re at the next danger of breast most cancers, which might positively enhance screening outcomes. A mix of density and danger evaluation approaches may very well be more practical in population-based screening applications for breast most cancers.

Journal reference:

  • Eriksson, M., Roman, M., Grawingholt, A., et al. (2023) European validation of an image-derived AI-based short-term danger mannequin for individualized breast most cancers screening—a nested case-control examine. The Lancet Regional Well being. doi:



Please enter your comment!
Please enter your name here

Most Popular

Recent Comments