Medical image processor and storage medium

1. A medical image processor comprising:

an input unit to input a cell shape image showing a shape of a cell in a tissue slice and a fluorescent image showing expression of a specific protein in a same range of the tissue slice as a fluorescent bright point;

a cell nucleus extracting unit which extracts a region of a cell nucleus from the cell shape image;

a fluorescent bright point extracting unit which extracts a fluorescent bright point from the fluorescent image;

a region estimating unit which sets a predetermined region which includes from a centroid of a region of each cell nucleus extracted by the cell nucleus extracting unit to an expression region of the specific protein in a cell membrane periphery, wherein when the set region does not overlap with another set region, the set region is estimated to be a cell region including one cell, and when a plurality of the set regions overlap with each other, a region surrounded by an outer circumference of the plurality of overlapping regions is estimated to be a cell region including a plurality of cells;

a feature amount calculating unit which calculates a feature amount regarding each cell region based on a cell nucleus and a fluorescent bright point included in each cell region estimated by the region estimating unit;

a determining unit which determines whether each cell region estimated by the region estimating unit is cancer and which determines an expression status of the specific protein in the region determined to be cancer, the determinations being based on the calculated feature amount; and

an output unit which outputs a determination result by the determining unit.

2. The medical image processor of claim 1 , wherein, the region set by the region estimating unit is a circular region with a centroid of the region of each cell nucleus extracted by the cell nucleus extracting unit as a center and a predetermined radius including the expression region of the specific protein in the cell membrane periphery.

3. The medical image processor of claim 1 , wherein,

the feature amount calculating unit calculates an area which each cell nucleus occupies in each cell region and/or an index value which shows uneven distribution of the cell nucleus in each cell region based on the cell nucleus included in each cell region estimated by the region estimating unit and calculates a density of the fluorescent bright point in each cell region based on the fluorescent bright point included in each cell region; and

the determining unit determines whether each cell region is cancer based on the area which each cell nucleus occupies in each cell region and/or the index value which shows uneven distribution of the cell nucleus in each cell region, and determines the expression status of the specific protein in the cell region determined to be cancer based on the density of the fluorescent bright point in the cell region.

4. The medical image processor of claim 1 , wherein,

the determining unit classifies the expression status of the specific protein in the cell region determined to be cancer into a plurality of levels based on whether the calculated feature amount of the fluorescent bright point in the cell region determined to be cancer exceeds a plurality of predetermined thresholds; and

the output unit outputs an image dividing the cell region determined to be cancer on the cell shape image in a form according to a classification result of the expression status of the specific protein as the determination result.

5. A non-transitory computer-readable storage medium including a program for controlling a computer to function as:

an input unit to input a cell shape image showing a shape of a cell in a tissue slice and a fluorescent image showing expression of a specific protein in a same range of the tissue slice as a fluorescent bright point;

a cell nucleus extracting unit which extracts a region of a cell nucleus from the cell shape image;

a fluorescent bright point extracting unit which extracts a fluorescent bright point from the fluorescent image;

a region estimating unit which sets a predetermined region which includes from a centroid of a region of each cell nucleus extracted by the cell nucleus extracting unit to an expression region of the specific protein in a cell membrane periphery, wherein when the set region does not overlap with another set region, the set region is estimated to be a cell region including one cell, and when a plurality of the set regions overlap with each other, a region surrounded by an outer circumference of the plurality of overlapping regions is estimated to be a cell region including a plurality of cells;

a feature amount calculating unit which calculates a feature amount regarding each cell region based on a cell nucleus and a fluorescent bright point included in each cell region estimated by the region estimating unit;

a determining unit which determines whether each cell region estimated by the region estimating unit is cancer and which determines an expression status of the specific protein in the region determined to be cancer, the determinations being based on the calculated feature amount; and

an output unit which outputs a determination result by the determining unit.