Qinetiq Limited

Qinetio Limited

Margaret Jai Varga

Paul Gerard Ducksbury

Maria Petrou

Christelle Marie Guittet

Anastasios Kesidis

Roberto Cipolla

Michael J Hamer

A method for automated selection of image regions ('tiles') in an overall image includes computing a gray-level co-occurrence matrix and entropy of a green plane of the overall image, applying morphological closing to the matrix and thresholding the matrix and entropy image to provide binary masks. The matrix and entropy masks are combined with a vignette mask, the combination indicating areas of acceptable tissue from which tiles are selected randomly. For cancer grading; image data is transformed to Hue, Saturation and Value; for steroid/protein expression analysis it is transformed to cyan and a Sobel of cyan is computed. A feature measure is computed for each tile based on color and texture, and is carried out randomly but influenced by feature measure. Finally, from the further selection, tiles are chosen which combine high feature measure with low overlap.

Automated selection of image regions

A method of grading tubules in a first histological slide image derives a second image of objects in the first image of objects in the first image with boundary characteristics corresponding to tubules. It also derives a third image of second objects in the first image having pixel value characteristic of fat and holes within tubules. It combines data from the second and third images to identify holes within tubules and determines the relative areas of holes as proportions of their tubules to provide ratios, individual tubule ratios and an overall ration for all holes and tubules collectively. The number of tubules containing appreciably sized holes is counted. Tubules are graded by thresholding based on individual and overall tubule/hole area ratios, tubule/object proportion, tubule number and number of tubule with appreciably sized holes. Thresholds are derived from image gradation by an appropriate medical expert.

Automated histological grading of tubules

A method of scoring Oestrogen and Progesterone Receptors expression (ER and PR) from histological images determines the percentage of brown image blob area in total blob area and derives percentage thresholds to quantify scoring. Brown blob area proportion is then compared with the thresholds to provide a first contribution to scoring of ER or PR. The number of relatively dark pixels is then counted, pixel number thresholds are determined to quantify scoring, and the number of relatively dark pixels is compared with the thresholds to provide a second contribution scoring of ER or PR. The two contributions are then added to provide an overall scoring in the range 0 to 8 which may be taken as a final score or it may be scaled to a conventional range 0 to 3.

Scoring estrogen and progesterone receptors expression based on image analysis

A method of histological assessment of nuclear pleomorphism to identify potential cell nuclei divides image data into overlapping sub-images. It uses principal component analysis to derive monochromatic image data, followed by Otsu thresholding to produce a binary image. It removes image regions at sub-image boundaries, unsuitably small image regions and holes in relatively large image regions. It then reassembles the resulting sub-images into a single image. Perimeters (P) and areas (A) of image regions which are potential cell nuclei are determined and used in calculating nuclear shape factors P/A. Nuclear pleomorphism is assessed as relatively low, moderate or high according to whether predetermined shape factor thresholds indicate a mean cell nucleus shape factor for an image is relatively low, moderate or high.

Histological assessment of nuclear pleomorphism

A method of measurement of mitotic activity from histopathological specimen images initially identifies image pixels with luminances corresponding to mitotic figures and selects from them a reference pixel to provide a reference color. Pixels similar to the reference color are located; image regions are grown on located pixels by adding pixels satisfying thresholds of differences to background and image region luminances. Grown regions are thresholded in area, compactness, width/height ratio, luminance ratio to background and difference between areas grown with perturbed thresholds. Grown regions are counted as indicating mitotic figures by thresholding region number, area and luminance. An alternative method of measuring mitotic activity measures a profile of an image region and counts the image region as corresponding to a mitotic figure if its profile is above a threshold at an intensity associated with mitotic figures. A mitotic figure is also indicated if the profile does not meet the previous criterion but has three other values satisfying respective threshold criteria.

Measurement of mitotic activity

The invention describes a wavelet compression method for color converted image planes (Y, Cb and Cr), wherein a relatively low (e.g. zero) degree of compression is performed in areas of an original color image having relatively higher importance, a relatively higher degree of compression in those areas indicated to be of lower importance. It performs a hierarchical encoding of a reduced wavelet image by discarding wavelet coefficients which satisfy the two criteria of firstly corresponding to image areas of relatively lower importance and secondly being below a certain wavelet coefficient threshold. The wavelet coefficient threshold is determined from a calculation histogram to remove a percentage of the image and can be specified as an input parameter by a user.

Data compression for colour images using wavelet transform

A method of measuring oestrogen or progesterone receptor (ER or PR) comprises identifying in histopathological specimen image data pixel groups indicating cell nuclei, and deriving image hue and saturation. The image is thresholded using hue and saturation and preferentially stained cells identified. ER or PR status is determined from normalised average saturation and proportion of preferentially stained cells. A method of measuring C-erb-2 comprises correlating window functions with pixel sub-groups to identify cell boundaries, computing measures of cell boundary brightness and sharpness and brightness extent around cell boundaries, and comparing the measures with comparison images associated with different values of C-erb-2. A C-erb-2 value associated with a comparison image having similar brightness-related measures is assigned. A method of measuring vascularity comprises deriving image hue and saturation, producing a segmented image by hue and saturation thresholding and identifying contiguous pixels. Vascularity is determined from contiguous pixel area corresponding to vascularity expressed as a proportion of total image area.

Growing community of inventors

Malvern, United Kingdom

Margaret Jai Varga