A neural network for identifying papillary carcinoma

Proof that machines 'think' but in their own matrix-multiplicative manner. Full text at http://www.jpathinformatics.org/text.asp?2018/9/1/43/246760.

Steps of analysis

Final classification

Hepatitis C with cirrhosis

Cirrhosis in a patient with Hepatitis C

CT

FNAC - ? hepatocytes

Cirrhotic nodules with dense lymphoid infiltrate

Apoptotic cells

Steatosis

Lung cytology

Few findings from guided lung FNACs.

Normal bronchial epithelium

Small cell carcinoma

Note the DNA spread over in thin strands

... pointing to a small cell carcinoma

Squamous carcinoma

An old smoker. No better place for a squamous carcinoma to develop.

or is it?

There's the blue cytoplasm

and a few tadpole cells

Few salivary lesions

Adenoid cystic carcinoma

'cribriform'

A nicer looking version

Acinic cell carcinoma

Cytology

Granular cytoplasm

Polymorphic adenoma

Looking like fibroadenoma, with plasmacytoid myoepithelial cells

More plasmacytoid cells

Stromal fragment

Pleomorphic adenoma on cytology

Epithelial myoepithelial carcinoma

Note the clear cells around acini

Warthin tumor

Note the double layered oncocytic epithelium

Basal cell adenoma

Mucoepidermoid cancer in a teenage girl

Mixed tumor on cytology

Note bright red to purple stromal fragments

Polymorphic adenoma, ? canalicular adenoma