NEW YORK – Having secured the first US Food and Drug Administration clearance of a digital cytology platform, Hologic is preparing to launch in the US its platform that combines algorithm-driven recognition of potentially cancerous abnormalities with composite images that give cytologists a high-resolution view of the cells layered throughout a sample.
Last week, the US Food and Drug Administration granted de novo marketing authorization for Hologic's Genius Digital Diagnostics System, which the company expects to launch in the US early this year. While the Marlborough, Massachusetts-based firm said it is the first clearance issued by the agency for a "digital cytology system that combines deep-learning-based artificial intelligence (AI) with advanced volumetric imaging technology to help identify precancerous lesions and cervical cancer cells," the Genius has been available in Europe for about two years and is sold also in Australia and New Zealand.
According to Hologic, the instrument improves on the sensitivity of microscopy-based examination of Pap test samples, and in testing, use of the system resulted in a 28 percent reduction of false negatives of high-grade squamous intraepithelial lesions and more severe lesions compared to microscopic review. The system is comprised of a digital imager, an artificial intelligence-developed algorithm for image analysis, an image management server, and a review station for examining patient samples.
"It can help more accurately detect cervical cancer, improve cytology workflow, and ultimately enhance patient care," Hologic CEO Stephen MacMillan said during the company's fiscal Q1 conference call last week.
MacMillan said that reading cytology slides has been a manual and labor-intensive process. In one of his first meetings with Quest Diagnostics about 10 years ago, he heard from the company that cytology workflows were among Quest's biggest concerns.
Michael Quick, VP of R&D and innovation for Hologic, said in a recent interview that his team developed the Genius system to be a next-generation cervical cancer screening tool. While surgical pathology has led the way in digitization of slide sample examination and adoption of AI-driven digitization tools, Quick said that cytology had lagged on technology for analysis of cellular samples.
He said the firm applied the lessons of about two decades of cervical cancer screening into developing the training data for its algorithm-backed software, Genius Cervical AI, which analyzes each of the approximately 80,000 cells in each sample.
The Genius system uses Hologic's existing ThinPrep Pap technologies and consumables for preparing slides for staining and evaluation, but Quick said Hologic expects that the Genius system will gradually replace its existing ThinPrep computer-assisted imaging and slide review system. That transition has been happening in Europe since the firm began commercializing the system there.
In April 2023, researchers from Hologic and collaborators in Germany published in Cancer Cytopathology study results that indicate the Genius system performed with higher overall sensitivity and a higher negative predictive value than the ThinPrep Imaging System for the detection of higher-grade and histologically confirmed cytologic abnormalities, with comparable specificity. The authors wrote that the systems were in concordance in about 87 percent of cases when using the same cytology categories, and the degree of concordance increased when the authors factored in histology results and further retrospective review of the cases. They recorded in only 0.65 percent of cases a major discrepancy such as a difference of two grades of cytology or a low-grade squamous intraepithelial lesion/high-grade squamous intraepithelial lesion shift.
In their retrospective study, the authors digitized and analyzed 1,992 cytology samples that had been evaluated in 2020 using the ThinPrep Imaging System as part of a routine screening program in Germany. The researchers also wrote that analysis required about 45 seconds per slide using the Genius system compared to 90 seconds per slide using the ThinPrep Imaging System.
Hologic isn't alone in its push to advance cytology's use of image recognition software. The firm's chief competitor in cytology, Becton Dickinson, also said last week that it had inked a collaboration deal with AI developer TechCyte to create an automated digital cervical cytology system for Pap testing. BD and TechCyte said they have developed a platform that is compatible with most liquid-based cytology preparations, including the BD SurePath Liquid-based Pap Test vial, and the firms intend to make it compatible with some of the most popular whole-slide imagers on the market.
BD officials also noted in their announcement last week that labs are dealing with shortages of skilled technicians, particularly in cytology.
Quick said that Hologic brings to the table technologies that allow for "volumetric" imaging "to scan in three dimensions a cytology sample to get an accurate representation of a high-quality, high-resolution image for cytology." The company's new cytology system captures images at different focal depths for each slide and creates a combined image in which every cell is in focus.
"We're able to capture the full depth of that volume of image and then say, 'Where is that particular cell in focus?'" he said. "We select that plane, and then we combine that with the other images that may be in focus in a different plane."
Quick, who is also a cytologist, said the Genius system's software helps cytologists more efficiently identify the most important information on each slide compared to microscopy.
The Genius system creates for each slide a gallery of 30 tiles. Those tiles include images that are representative of the cells in the sample and cells that the software has flagged for cytologists as suspicious and worthy of review. Quick said Hologic populates those galleries with diagnostically relevant objects of interest as well as normal but morphologically different cells.
The authors of the Cancer Cytopathology article noted that the Genius system's scanning method simultaneously captures data up to 14 focal planes to merge in-focus images from multiple planes, eliminating the need to scan individual layers within the sample. The gallery organizes cells into groups, with the cells with low nuclear-to-cytoplasmic ratio, enlarged nuclei, or koilocytes in one row, glandular cells and cell clusters in another, and microorganisms and infections in yet another.
When a cytologist selects an image from the gallery, the software zooms in on that area of the whole-slide image so they can see the surrounding cells and the context of any abnormalities, Quick said. The Genius software helps cytologists spend more time interpreting cells as diagnosticians and less time hunting for unusual cells, he said.
Quick said customers in Europe, particularly large commercial clinical laboratories, have been quick to adopt the Genius platform. More than 60 laboratories in Europe have installed the Genius system, and it is now being used in almost every major market on the continent including labs in the UK, France, Germany, Belgium, Switzerland, and the Nordic countries.
"We're seeing increases in disease detection that have been reported by laboratories out of Germany and Luxembourg, and we've seen increases in efficiency," he said. "Where this is important is that with the shortage of cytologists and pathologists, there's ever-demanding pressure on laboratories, and with the Genius system, they've seen about a doubling of their productivity."
Hologic is seeing year-over year growth in installations in Europe, he added.
Quick said cytologists have also expressed appreciation that the computer-assisted analysis technology is there to help them rather than replace them.
During Hologic's fiscal Q1, it recorded a 5 percent decline in its cytology and perinatal testing revenues in the quarter ended Dec. 30, to $120.0 million from $126.8 million a year earlier, which the company attributed to buildup of laboratory inventories during the second half of fiscal 2023. However, the firm previously reported in November that the firm had seen increased revenues in fiscal year 2023 year over year from its ThinPrep Pap Test products as more patients returned to clinics for wellness visits following a drop-off in visits earlier in the COVID-19 pandemic.