Room4 Group Limited

Institute for Medical Informatics

Oxford Cancer Biomarkers Ltd

Havard E Danielsen

John Robert Maddison

John Maddison

David Kerr

Birgitte Nielsen

An apparatus and computer-implemented method for training a machine-learning algorithm to perform histopathological analysis is disclosed. The method comprises obtaining () a plurality of first microscopic images of first histological specimens that have been stained with a first marker; and obtaining (), a respective plurality of second microscopic images of second histological specimens that have been stained with a second, different marker. The method further comprises obtaining () a respective plurality of mask images generated for the second microscopic images, each mask image identifying a histological feature of interest highlighted in the respective second microscopic image by the second marker. The method comprises training () the machine-learning algorithm to predict, from a first microscopic image, a histological feature of interest that would be highlighted in the same specimen by the second marker. Also disclosed is an apparatus and computer-implemented method for histopathological analysis using the trained machine-learning algorithm.

Histopathological image analysis

Quantitative analysis of a tissue sample includes carrying out a ploidy measurement on a plurality of nuclei of the sample to determine the ploidy type of a sample and also carrying out a stroma measurement on a section to determine the stroma type of a sample. The ploidy type, e.g. diploid or non-diploid, and the stroma type, high or low stroma, give an improved patient survival estimate than the ploidy type alone.

Tissue sample analysis technique

A computer-implemented segmentation method is used to process an image representing a plurality of nuclei. The method is implemented in a computer having a processor and a physical memory. A set of instructions are provided to the processor the physical memory of the computer. The processor is configured by executing the set of instructions in the physical memory so as to automatically segment the image by: thresholding a grey-scale image to create a black and white image; identifying objects in the black and white image and removing objects failing to meet predetermined criteria; extracting objects; and applying an edge detector on the segmented image to identify the edges of the nuclei. Overlapping nuclei are split to improve results.

Optimizing the initialization and convergence of active contours for segmentation of cell nuclei in histological sections

A method and apparatus for the analysis of cell nuclei may use a system where images captured from a bright field microscope are analysed and information is retrieved that give quantitative information of the different type of chromatin within the cell nuclei for example Hetrochromatin and Euchromatin. The method is based on segmented cell nuclei. Quantitative information is determined related to the greyscale distribution of the different areas of the chromatin types and the object structure within these different areas.

Microscopy analysis technique

A method and apparatus for aligning microscope images. Microscope images of the same or very similar subjects provided by different microscopes are aligned. The images from two types of microscope, such as a Virtual Microscope (VM) and a Light Microscope (LM) are used. An image produced by the virtual microscope is easily read by a viewer, as it represents a scan of a whole slide rather than individual high power fields of view. An area of the image can be selected for further examination or objective analysis by the LM microscope. The qualitative or quantitative information obtained from the Light microscope using the method described may then be located back into the virtual microscope image to provide understandable context.

Growing community of inventors

Tofte, Norway

Havard E Danielsen