In the quiet, high-stakes minutes after a CT scan is taken, a patient's fate is often decided by a radiologist's weary eyes and a ruler on a screen. Nucleo, a San Francisco startup founded in 2025, is betting that the critical first read of a cancer patient's scan should be handled by a consistent, near-instantaneous AI. The company's software ingests a CT scan and, in seconds, returns a detailed analysis of tumor lesions and body composition, a foundational step in oncology care that its founders claim is 2,500 times faster than manual methods and shows 98% agreement with medical experts [Y Combinator].
For Angelica Iacovelli and Luca Pegolotti, the co-founders, the goal is to build a reliable infrastructure layer beneath the clinician. The platform automates three specific, protocol-driven tasks: measuring fat and muscle mass for sarcopenia assessment, sizing tumor lesions, and classifying those lesions as target or non-target under the standardized RECIST criteria [Nucleo homepage].
The Clinical Wedge: RECIST, Sarcopenia, and Speed
Nucleo's product strategy targets measurement-intensive workflows where automation offers immediate value. In oncology, treatment decisions and clinical trial eligibility often hinge on precise, consistent tracking of tumor size changes over time. Manual segmentation and measurement are slow and prone to inter-observer variability. Similarly, assessing a patient's muscle mass (sarcopenia) from a CT scan is a strong prognostic indicator for chemotherapy tolerance, but it is rarely done in routine practice due to the time required.
Nucleo's reported validation metrics speak directly to these pain points. The claim of 98% agreement with expert reads addresses the paramount concern of accuracy in a regulated field [Y Combinator]. The 2,500x speed improvement translates to a scan analysis that takes seconds instead of hours, potentially collapsing the timeline from imaging to treatment planning [Y Combinator].
Founders with a Foot in Two Worlds
The company's technical credibility is anchored by the backgrounds of its two founders:
- Angelica Iacovelli, the CEO, brings experience from Stanford and a track record that earned a spot on a Forbes profile list for YC founders [Forbes, 2026].
- Luca Pegolotti, the CTO, was a postdoctoral researcher in Stanford's Cardiovascular Imaging Research Group before joining Apple's Health AI team in Zürich [Luca Pegolotti personal website, 2026].
Early Traction in Prestigious Hallways
For a company that only launched in Y Combinator's Fall 2025 batch, Nucleo has secured a roster of early clinical partners. The startup reports its tools are already in use at institutions including Stanford Hospital, Cedars-Sinai, Weill Cornell Medicine, and UCI Health [Y Combinator]. These are deployments within leading U.S. academic medical centers known for their rigorous vetting of new technologies. The company is also generating revenue, according to its Product Hunt page, indicating these hospital relationships have progressed beyond research collaborations to commercial deployments [Product Hunt].
The Road to Regulation and Reimbursement
The path forward is paved with the complex realities of medical technology. While the 98% agreement statistic is compelling, the gold standard for adoption is regulatory clearance from bodies like the U.S. Food and Drug Administration. Nucleo's current tools may be marketed as clinical decision support software, but any claim that moves toward autonomous diagnosis would likely trigger an FDA review process.
- Algorithmic drift. The performance of AI models can degrade over time or behave unexpectedly on scans from new hospital imaging equipment.
- Reimbursement motion. For widespread hospital adoption, Nucleo must prove its tool improves care metrics in a way that justifies its cost.
- Clinical workflow fit. Deep integration with major picture archiving and communication systems (PACS) is essential.
What Standard Care Looks Like Today
For a patient with a solid tumor, today's standard of care involves a radiologist manually scrolling through hundreds of CT scan slices, using software calipers to measure select lesions, and visually estimating body composition. This process is slow, subjective, and often inconsistently applied. Nucleo's bet is that by automating this quantitative foundation, it can bring a new level of speed, consistency, and depth to the care of every cancer patient facing a CT scan.