Nucleo

What do they actually do

Nucleo builds an AI tool that analyzes CT scans for oncology. Today it automatically measures body composition (e.g., sarcopenia metrics), segments and sizes tumor lesions, and classifies lesions as “target” vs “non‑target” to support RECIST‑style response tracking. The typical workflow is import CT → automated analysis → results and a report in seconds for use in clinical reporting or trial reads (Nucleo homepage • YC company page).

They are running pilots/validation with major centers (e.g., Stanford Hospital, Cedars‑Sinai, UCI Health, Weill Cornell) and publicly emphasize performance comparisons and expert‑agreement metrics, but do not list production‑grade PACS/EHR integrations or regulatory clearances on their site—consistent with a pilot‑stage product (YC company page • Nucleo homepage).

Who are their target customer(s)

• Medical oncologists at hospitals: They need fast, reliable tumor measurements and body‑composition metrics to select and dose therapy. Manual measurements vary by reader and slow decisions, pushing out treatment start dates (Nucleo homepage • YC company page).
• Diagnostic radiologists: Manual segmentation and RECIST tracking are time‑consuming and inconsistent. They need automation that reduces time per CT while preserving reproducibility for reports and tumor boards (Nucleo homepage • YC company page).
• Clinical trial imaging teams and researchers: Trials require standardized lesion measurements and RECIST‑style classifications at scale; manual reads increase variability, timelines, and cost. Teams need consistent, automatable endpoints and auditability for multi‑site studies (Nucleo homepage • YC company page).
• Hospital imaging/IT and operations leaders: They must lower throughput delays and staff burden while meeting security, integration, and QA requirements. Many early tools require manual uploads and lack PACS/EHR integration or clear regulatory status, making routine deployment hard (Nucleo homepage • YC company page).
• Nurse navigators / tumor‑board coordinators: Care coordination is disrupted when imaging interpretation is slow or inconsistent, causing appointment delays and rescheduling across disciplines. Faster, standardized outputs would improve timeline reliability (YC company page).

How would they acquire their first 10, 50, and 100 customers

• First 10: Run tightly scoped pilots with leading academic cancer centers (leveraging named partners as references), with hands‑on support, pre‑defined validation protocols, and commitments to co‑present results to create internal champions (YC company page).
• First 50: Standardize onboarding and offer a simple DICOM connector to eliminate manual uploads; convert pilots into paid proof‑of‑value across regional cancer centers and trial groups using published validations, security packets, and 1–2 clinical sales plus a medical‑affairs lead.
• First 100: Scale through partnerships with imaging vendors/CROs and oncology software providers, while productizing self‑serve deployment/monitoring/compliance. Progress regulatory and QA to reduce procurement friction and add customer success for multi‑site rollouts.

What is the rough total addressable market

Top-down context:

AI in medical imaging is an estimated $1.3–1.7B global market in 2024–2025 and growing rapidly; CT is the largest modality by use, which aligns with Nucleo’s focus (Grand View Research).

Bottom-up calculation:

US oncology use case: ~2.04M new cancer cases projected in 2025. If 80% receive CT‑based staging/response imaging and average 4 CT scans in year one, that’s ~6.5M oncology CTs. At $25 per analyzed scan, US TAM ≈ $160M/year; pricing or scan counts could raise this materially (ACS Cancer Statistics 2025).

Assumptions:

• 80% of new cancer cases undergo CT‑based imaging suitable for automated metrics in the first year of care.
• Average 4 CT scans per patient in year one (staging + early response).
• Average net realized price of ~$25 per scan for automated lesion/body‑composition analysis; excludes multi‑year follow‑up and international markets.

Who are some of their notable competitors

• Mint Medical (MINT Lesion): Structured oncology reads with built‑in RECIST/iRECIST workflows used in clinical trials; automates response classification and longitudinal lesion tracking—established in sites and CROs (Mint Medical • RECIST resources).
• TeraRecon (Intuition): Enterprise advanced visualization with lesion segmentation, volumetrics, and support for tracking using RECIST/WHO comparisons; widely deployed and integrated into hospital workflows (TeraRecon Reg. Resource—RECIST tracking • Intuition brochure).
• Quibim: Radiomics/AI platform for oncology with automated organ/lesion segmentation and imaging biomarker panels across CT/MRI/PET; used for trial biomarker discovery and standardized quantification (Quibim Oncology • Products).
• AIQ Solutions (TRAQinform IQ): Automates lesion‑level detection and longitudinal response assessment across all disease burden; published evidence of prognostic value vs. RECIST/PERCIST in retrospective cohorts (AIQ site • EJNMMI study).
• Median Technologies (iCRO, iSee): Oncology clinical‑trial imaging CRO with software (iSee) for lesion identification/tracking and RECIST/iRECIST operations, plus AI Imaging Lab for predictive biomarkers—deep experience in multi‑site trials (Median iCRO • iSee Viewer).