AI Device Achieves 97% Accuracy in Skin Cancer Detection

PLUS: AI identifies and classifies pancreatic lesions by 99.9%

Hey there

Sorry for the late issue of the Cancer Innovations newsletter this week. I had a sick child and had to focus on her instead of researching and writing this.

If you have any questions, feel free to hit reply. You can also find me on Twitter @thomassorheim 

- Thomas

Today’s topics 👇️ 

  1. Non-Invasive Device Achieves 97% Accuracy in Skin Cancer Detection

  2. The Next-Gen Robotic Arm powered by AI for HIFU

  3. AI Predicts Lung Cancer in Nonsmokers with Startling Accuracy

  4. AI Tool PANDA Revolutionizes Early Detection of Pancreatic Cancer

Non-Invasive Device Achieves 97% Accuracy in Skin Cancer Detection

Great news in skin cancer detection! A groundbreaking AI-driven device is reshaping the future of skin cancer diagnosis. The DermaSensor boasts an impressive 97% success rate and matches top dermatologists' diagnostic acumen. This innovative, non-invasive tool empowers primary care physicians to detect various skin cancers with unparalleled ease and efficiency. A triumph in Australia, the DermaSensor is eagerly anticipated in the USA, awaiting FDA approval. Soon, dermatologists and oncologists should have another life-saving technology to fight skin cancer. 

Link to research paper: ScienceDirect
News coverage: Yahoo
Doctors: Dr. Armand Cognetta
Company & AI: DermaSensor 
Institutions mentioned: FSU College of Medicine (Florida State University College of Medicine)

The Next-Gen Robotic Arm powered by AI for HIFU

Researchers at the University of Waterloo have developed a groundbreaking robotic arm powered by AI to transform cancer treatment. This pioneering device precisely targets tumors with High-Intensity Focused Ultrasound (HIFU), significantly reducing damage to healthy cells and cutting recovery time from ten days to two. It is the first of its kind for non-invasive cancer therapy and marks a major leap forward. The arm is currently progressing toward FDA approval. 

News coverage: CBC News
Doctor’s mentioned: Moslem Sadeghi Goughari
Institutions mentioned: University of Waterloo 

AI Predicts Lung Cancer in Nonsmokers with Startling Accuracy

The "CXR-Lung-Risk" AI model, created by Anika Walia and her team at Boston University and Harvard Medical School, represents a breakthrough in detecting high-risk lung cancer among nonsmokers. Trained on a substantial dataset of 147,497 historical chest X-rays from 40,643 asymptomatic individuals, this model is adept at identifying those at elevated risk. In a study with 17,407 patients, the AI accurately classified 28% as high risk, with 2.9% of these later confirmed to have lung cancer. Crucially, the high-risk group was 2.1 times more likely to develop lung cancer compared to the low-risk group. This method of using past X-ray data to predict future cancer incidence offers a novel and effective approach to lung cancer screening, especially vital for nonsmokers, enhancing early detection and potentially improving treatment outcomes.

News coverage: MedicalExpress
Doctor’s mentioned:

Institutions mentioned: 

AI Tool PANDA Revolutionizes Early Detection of Pancreatic Cancer

The recent study introduces PANDA, an AI tool utilizing non-contrast CT scans for detecting pancreatic ductal adenocarcinoma (PDAC) early, a cancer notorious for high mortality due to typically late discovery. This international collaboration involving China, the USA, and the Czech Republic highlights PANDA's exceptional ability to identify and classify pancreatic lesions, surpassing traditional methods accurately. It outperforms radiologists in detecting lesions with 92.9% sensitivity and 99.9% specificity, distinguishing between eight pancreatic lesion subtypes, including PDAC. This breakthrough offers a potential paradigm shift in cancer screening, significantly improving early diagnosis and treatment outcomes for PDAC.

Link to research paper: Nature Medicine
News coverage: News Medical
Doctor’s:

  • Kai Cao

  • Yingda Xia

  • Jiawen Yao

  • Xu Han

  • Lukas Lambert

  • Tingting Zhang

  • Wei Tang

  • …and more (for full list, see research paper)