How the AI Deciphers CT Scans
Altis Labs, a San Francisco‑based startup, closed a $25 million Series A round on September 21, 2026. The funding will support the company’s artificial‑intelligence platform that reads computed‑tomography (CT) scans from oncology clinical trials and predicts patient survival. Investors include prominent venture funds that specialize in medical technology. Altis Labs plans to use the capital to scale its software, expand its data set, and accelerate regulatory approval.
Breaking news
How Predictive Diagnostics Are Reshaping the Automotive Insurance Ecosystem
Preparing for the EU AI Act’s Strict Transparency Mandates
Widespread Firebase SDK Failure Triggers Massive iOS App Crashes
Former Tesla Engineers Secure Funding for Logistics AutomationThe core of Altis Labs’ technology is a deep‑learning model trained on thousands of CT images linked to detailed clinical outcomes. By learning patterns in tumor size, shape, and density, the AI produces a risk score that correlates with how long a patient is likely to live after treatment. The algorithm can also flag subtle changes that may indicate early disease progression, offering clinicians a quantitative tool to guide therapy decisions. The startup’s founders claim the model outperforms traditional radiologic assessment in several early‑stage studies.
Altis Labs’ system first segments the scan into individual organs and lesions. It then extracts quantitative features such as volume, attenuation, and texture. These metrics feed into a neural network that has been calibrated against clinical endpoints. The model outputs a probability of survival at 12, 24, and 36 months. In a validation cohort, the AI’s predictions matched actual outcomes within a 5‑percentage‑point margin, surpassing conventional radiology reports. The company’s engineers emphasize that the model is transparent; clinicians can view the contributing features that drive each prediction.
What This Means for Oncology Trials
By providing objective imaging biomarkers, Altis Labs aims to streamline trial design. Sponsors can use the AI score as an early surrogate endpoint, potentially shortening study duration. The platform also enables adaptive randomization: patients with poorer predicted survival may be steered toward experimental therapies. Regulatory agencies are increasingly receptive to data‑driven endpoints, so the startup’s tool could facilitate faster drug approvals. Moreover, patients benefit from more precise prognostication, allowing them to make informed choices about treatment intensity.
The $25 million infusion will fund the expansion of the training data set to include diverse populations and imaging modalities beyond CT, such as positron‑emission tomography. Altis Labs also plans to develop a cloud‑based interface that integrates with hospital picture archiving and communication systems. The company’s roadmap includes filing for an FDA clearance of its software as a medical device by 2028.
In the long term, Altis Labs envisions its AI becoming a standard component of oncology care. If widely adopted, the technology could reduce variability in imaging interpretation and improve survival outcomes across multiple cancer types. The company’s success will hinge on continued validation, regulatory clearance, and partnerships with pharmaceutical sponsors.
Frequently Asked Questions
What data does Altis Labs use to train its AI? The model is trained on thousands of anonymized CT scans from completed and ongoing oncology trials, paired with patient survival data and clinical annotations.
Will the AI replace radiologists? No. Altis Labs positions its tool as a decision support system that augments radiologists’ expertise, not a replacement.
When can patients expect to see this technology in clinical practice? After the company secures FDA clearance, which it aims to achieve by 2028, the software could be integrated into trial protocols and eventually routine care.
