9 Comments

Amazing stuff. Even without AI, I have increasingly received human interpreted radiology reports in which calcifications and atherosclerosis are reported as incidental findings on x-rays and CT scans, which for me prompts a more aggressive conversation about risk reduction with my patients.

This is all rolling out so fast it will be difficult for physicians to keep up, we need a central AI continuing medical education resource. Are there any? (In addition to you Dr Topol 😉)

Expand full comment
author

Not easy. Trying to keep up!

Expand full comment
founding
Mar 31Liked by Eric Topol

Just plain WOW from a data scientist.

Expand full comment
author

That's what I thought too.

Expand full comment
Mar 31Liked by Eric Topol

Impressive! I imagine that in a few years, simple, inexpensive, and commonly ordered tests like CXR, chest CT, ECG, and CBC could offer significantly more secondary analytical and predictive insights. And, it would be prudent to assume that this information will be interchangeable between modalities (proven noninferiority among modalities), and reviewing recent tests would be sufficient for assessing these risks rather than ordering new ones or using conventional risk tools.

Expand full comment
author

Agree! It's coming

Expand full comment
Mar 31Liked by Eric Topol

What good information! I'm going to send your article to my husband's cardiologist.

Expand full comment

This greatly expands my vision of what AI

can do to expand the uses for scans and other medical tests.. Fascinating. Thanks, Dr Topol.

Expand full comment
May 27·edited May 27

Thanks for covering this line of amazing work. We are building PANDA 2 for particularly enhancing the detection sensitivity on pancreatic tumors <2cm (T1). I lead the lab who build PANDA and PANDA+

a few more things to come (we keep transparency on what we build while extensively performing multi-center clinical validation and real world validation now):

1) Improved Esophageal Varices Assessment on Non-Contrast CT Scans, (early accept, Top 11%), MICCAI 2024;

2) Parse and Recall: Towards Accurate Lung Nodule Malignancy Prediction like Radiologists. MICCAI 2023 (early accept)

3) Cluster-Induced Mask Transformers for Effective Opportunistic Gastric Cancer Screening on Non-contrast CT Scans. MICCAI 2023;

"Effective Opportunistic Gastric Cancer Screening on Noncontrast CT Scans", Scientific Oral, RSNA 2023

4) Liver Tumor Screening and Diagnosis in CT with Pixel-Lesion-Patient Network. MICCAI 2023;

"Automatic Liver Tumor Screening and Differential Diagnosis in CT Using Pixel-Lesion-Patient Network with Reader Study and External Validation", Scientific Oral, RSNA 2023

5) Effective Opportunistic Esophageal Cancer Screening using Noncontrast CT Imaging, MICCAI 2022 (early accept);

"Effective Opportunistic Esophageal Cancer Screening Using Non-contrast CT Imaging (#9333)", Scientific Oral, RSNA 2022

6) Thoracic DeepCRC: Automatic Colon Rectum and Colorectal Cancer Segmentation in CT scans with Global Attention and Deep Coordinate Transform, MICCAI 2022 (early accept);

"Effective Opportunistic Screening for Colorectal Cancer using Abdominal or Chest Noncontrast CTs", Scientific Poster, RSNA 2023

"Detection of Colorectal Cancer in Regular Abdominal CT Scans without Bowel Preparation using Deep Learning", Scientific Oral, RSNA 2022

7) "Opportunistic Breast Cancer Screening using Non-contrast CT Imaging.", Scientific Poster, RSNA 2023

last but not least, 8) Effective Lymph Nodes Detection in CT Scans Using Location Debiased Query Selection and Contrastive Query Representation in Transformer, CoRR abs/2404.03819 (2024)

Expand full comment