Solutions

Experts to annotate
images at scale

90% of all healthcare data is medical images. Whether generated by smartphones or regulated medical imaging devices, this imaging data represents a significant opportunity for AI development.

Our annotation platform enables clients to structure medical images by both classifying images and segmenting regions of interest.

Talk to an expert LEARN MORE →

Use cases

Skin, face and waste images

Tag skin images for the presence and severity of lesions i.e. lacerations, psoriasis, acne or rashes. Classify skin tone for granular color matching. Classify and segment bodily wastes and fluids to determine interventions.

Ultrasound

Identify pleural and b-lines in lung ultrasounds. Segment areas of abnormal blood flow, fetal abnormalities, or gallstones.

X-ray

Identify the presence of a lesion, e.g. cavity, septal lines or broken bone. Segment the location of that lesion.

Pathology slide

Identify cellular processes e.g. mitosis, to determine mitotic rate of a cancer. Classify cells as high or low grade, to determine differentiation from tumor cells from healthy cells. Classify the presence of and segment other cellular or molecular features.

MRI, CT and PET

Identify the presence of a lesion, e.g. tumor, lung nodule, brain bleed, or area of decreased blood flow. Segment the location of that lesion.

Annotation types

Classification

Polygon segmentation

Box segmentation

Line segmentation

Circle segmentation

Testimonials

Trusted by AI leaders across healthcare

“We were able to improve our model dramatically - from .6 to .83 F1 score - in part, because of Centaur Labs.”

Fausto Milletarì

Sr. AI Scientist

Read the story →

"We’re excited by the high throughput and quality of annotations we get with Centaur Labs - I would definitely recommend working with them.”

Sarah Bettigole

Head of Immunology and Data Science

Read the story →

“We were able to use the disagreement between labelers to convert categorical labels into a continuous metric. That’s ultimately the training data we used and it was only possible because of the 20+ opinions we could gather from Centaur Labs on each piece of our data”.