Abstract: A phase image is formed by calculation from a hologram image of a cell, and segmentation is performed for each pixel for the phase image using a fully convolution neural network to identify an undifferentiated cell region, a deviated cell region, a foreign substance region, and the like. When learning, when a learning image included in a mini-batch is read, the image is randomly inverted vertically or horizontally and then is rotated by a random angle. A part that has been lost within the frame by the pre-rotation image is compensated for by a mirror-image inversion with an edge of a post-rotation image as an axis thereof. Learning of a fully convolution neural network is performed using the generated learning image. The same processing is repeated for all mini-batches, and the learning is repeated by a predetermined number of times while shuffling the training data allocated to the mini-batch. The precision of the learning model is thus improved.

Abstract: A training label image correction method includes performing a segmentation process on an input image (11) of training data (10) by a trained model (1) using the training data to create a determination label image (14), comparing labels of corresponding portions in the determination label image (14) and a training label image (12) with each other, and correcting label areas (13) included in the training label image based on label comparison results.

Abstract: A training label image correction method includes performing a segmentation process on an input image (11) of training data (10) by a trained model (1) using the training data to create a determination label image (14), comparing labels of corresponding portions in the determination label image (14) and a training label image (12) with each other, and correcting label areas (13) included in the training label image based on label comparison results.

Abstract: A phase image is formed by calculation from a hologram image of a cell, and segmentation is performed for each pixel for the phase image using a fully convolution neural network to identify an undifferentiated cell region, a deviated cell region, a foreign substance region, and the like. When learning, when a learning image included in a mini-batch is read, the image is randomly inverted vertically or horizontally and then is rotated by a random angle. A part that has been lost within the frame by the pre-rotation image is compensated for by a mirror-image inversion with an edge of a post-rotation image as an axis thereof. Learning of a fully convolution neural network is performed using the generated learning image. The same processing is repeated for all mini-batches, and the learning is repeated by a predetermined number of times while shuffling the training data allocated to the mini-batch. The precision of the learning model is thus improved.

Abstract: A shape of a cell is favorably observed in a non-invasive manner.

Abstract: Disclosed herein is a partially movable radiation tomography apparatus capable of superimposing an anatomical image and a functional image without causing misalignment. The apparatus of the disclosure includes a device associated with a CT gantry having a top board, and a device associated with a PET gantry. The latter device is attachable and removable to/from the former device. This apparatus can reduce positional misalignment between the images captured by these devices. The present disclosure involves making a correction for positioning the subject images captured by the devices associated with the respective gantries with respect to each other based on a calibration image on which markers m attached to those gantries are shot. Such a configuration allows for positioning the subject images with possible misalignment between the gantries taken into account, and eventually reducing positional misalignment between the subject images captured by the devices associated with the respective gantries.

Abstract: The disclosure has an object to provide a medical-data processing device that allows generation of an MIP image suitable for diagnosis. With the medical-data processing device of the disclosure, intensity of a body surface region in three-dimensional space data is adjusted. The body surface region corresponds to a body surface of a stereoscopic image of a subject. Since the intensity of the body surface region is adjusted to be decreased, the maximum intensity is selected from a portion except for the body surface region to generate the MIP image. This prevents the body surface of the subject from appearing upon generating the MIP image. Therefore, the MIP image is obtainable having excellent visibility to the inside of the subject.

Abstract: Disclosed herein is a partially movable radiation tomography apparatus capable of superimposing an anatomical image and a functional image without causing misalignment. The apparatus of the disclosure includes a device associated with a CT gantry having a top board, and a device associated with a PET gantry. The latter device is attachable and removable to/from the former device. This apparatus can reduce positional misalignment between the images captured by these devices. The present disclosure involves making a correction for positioning the subject images captured by the devices associated with the respective gantries with respect to each other based on a calibration image on which markers m attached to those gantries are shot. Such a configuration allows for positioning the subject images with possible misalignment between the gantries taken into account, and eventually reducing positional misalignment between the subject images captured by the devices associated with the respective gantries.

Abstract: A PET apparatus and a processing method according in this invention carry out arithmetic processes in parallel, in steps S4 (forward projection) and S5 (back projection), for every LOR, on list data (created from event data obtained by detecting gamma rays), and can therefore be applied to other arithmetic mechanisms and have versatility. Since parallel processing is carried out on the list data and the parallel processing is carried out in each of steps S4 (forward projection) and S5 (back projection), a competition for memory can be prevented to realize an improvement in speed. As a result, high versatility and an improvement in speed of the calculations can be attained.

Abstract: The disclosure has an object to provide a medical-data processing device that allows generation of an MIP image suitable for diagnosis. With the medical-data processing device of the disclosure, intensity of a body surface region in three-dimensional space data is adjusted. The body surface region corresponds to a body surface of a stereoscopic image of a subject. Since the intensity of the body surface region is adjusted to be decreased, the maximum intensity is selected from a portion except for the body surface region to generate the MIP image. This prevents the body surface of the subject from appearing upon generating the MIP image. Therefore, the MIP image is obtainable having excellent visibility to the inside of the subject.

Abstract: Provided is a data processor used for imaging a plurality of subjects collectively. In the data processor, trimming is automatically performed to spatial data containing data on three-dimensional information for enhancing working efficiencies of experiments. Specifically, data on the subjects contained in the spatial data is divided into individual divisional data automatically and collectively. This eliminates necessity for individual trimming of the cross sectional images by the experimenter, resulting in significantly facilitating the latter image analysis.

Abstract: A PET apparatus and a processing method according in this invention carry out arithmetic processes in parallel, in steps S4 (forward projection) and S5 (back projection), for every LOR, on list data (created from event data obtained by detecting gamma rays), and can therefore be applied to other arithmetic mechanisms and have versatility. Since parallel processing is carried out on the list data and the parallel processing is carried out in each of steps S4 (forward projection) and S5 (back projection), a competition for memory can be prevented to realize an improvement in speed. As a result, high versatility and an improvement in speed of the calculations can be attained.