Deep learning, a type of artificial intelligence (AI), was able to identify malignant lung nodules on low-dose chest computed tomography (LDCT) scans with a performance meeting or surpassing that of professional radiologists, reports a new research from Google and Northwestern Medicine.
Image credit: Northwestern Medicine
This deep-learning system offers an automated image evaluation system to improve the accuracy of early lung cancer diagnosis that could result in earlier treatment. The deep-learning system was compared against radiologists on LDCTs for patients, a few of whom had biopsy established cancer within a year. In the majority of comparisons, the model performed on par or better than radiologists.
Deep learning is a method that instructs computers to learn by example. The deep-learning system also generated fewer false positives and fewer false negatives, which could result in fewer pointless follow-up procedures and fewer overlooked tumors if it were used in a clinical environment.
The paper was published in the May 20
th issue of Nature Medicine.
“Radiologists generally examine hundreds of two-dimensional images or ‘slices’ in a single CT scan but this new machine learning system views the lungs in a huge, single three-dimensional image,” said research co-author Dr. Mozziyar Etemadi, a research assistant professor of anesthesiology at Northwestern University Feinberg School of Medicine and of engineering at McCormick School of Engineering. “AI in 3D can be much more sensitive in its ability to detect early lung cancer than the human eye looking at 2D images. This is technically ‘4D’ because it is not only looking at one CT scan, but two (the current and prior scan) over time.
“In order to build the AI to view the CTs in this way, you require an enormous computer system of Google-scale. The concept is novel but the actual engineering of it is also novel because of the scale.”
Etemadi heads his study team while also in anesthesiology residency training at Northwestern as part of an exclusive residency research track.
Etemadi’s twin roles enable research in his lab to traverse the technological and communications boundaries between engineering and healthcare. His lab is situated inside one of the intensive care units at Northwestern Memorial Hospital to enable continuous communication among engineers and nurses, physicians and other caregivers.
This area of research is incredibly important, as lung cancer has the highest rate of mortality among all cancers, and there are many challenges in the way of broad adoption of lung cancer screening. Our work examines ways AI can be used to improve the accuracy and optimize the screening process, in ways that could help with the implementation of screening programs. The results are promising, and we look forward to continuing our work with partners and peers.
Shravya Shetty, Technical Lead, Google
Lung cancer is the most common cause of cancer-linked fatalities in the United States, resulting in an estimated 160,000 deaths in 2018. Large clinical trials spanning the United States and Europe have revealed that chest screening can identify the cancer and lower death rates. However, high inaccuracy rates and the limited access to these screenings mean that a number of lung cancers are typically detected at advanced stages, when they are not easy to treat.
The deep-learning system makes use of the primary CT scan and, as well as whenever available, a prior CT scan from the patient as input. Prior CT scans are beneficial in predicting lung cancer malignancy possibility because the growth rate of suspicious lung nodules can be symptomatic of malignancy. The computer was taught using entirely de-identified, biopsy-confirmed low-dose chest CT scans.
The novel system identifies both a region of interest and if the region has a high probability of lung cancer.
The model outdid six radiologists when prior CT imaging was not present and did as well as the radiologists when there was prior imaging.
“The system can categorize a lesion with more specificity. Not only can we better diagnose someone with cancer, we can also say if someone doesn’t have cancer, potentially saving them from an invasive, costly and risky lung biopsy,” Etemadi said.
Google researchers created the deep-learning model and applied it to 2,763 de-identified CT scan sets offered by Northwestern Medicine to corroborate the new system’s accuracy. The researchers learned the AI-driven system was able to detect sometimes-tiny malignant lung nodules with a model AUC of 0.94 test cases. The cases were taken from the Northwestern Electronic Data Warehouse as well as other Northwestern Medicine data sources, as a result of complex, highly tailored software engineered by Etemadi’s team.
“Most of the software we use as clinicians is designed for patient care, not for research,” Etemadi said. “It took over a year of dedicated effort by my entire team to extract and prepare data to help with this exciting project. The ability to collaborate with world-class scientists at Google, using their unprecedented computing capabilities to create something with the potential to save tens of thousands of lives a year is truly a privilege.”
The researchers warn that these findings have to be clinically verified in large patient populations, but they say this model may help in refining the management and outcome of lung cancer patients.
The paper’s corresponding author is Dr. Daniel Tse, Google product manager.