Automatic determination method of at least one parameter indicative of the degree or level of functionality of a lung
A determination method of at least one parameter indicative of the functionality degree or level of a lung according to the ventilation rate of its cells is provided, on the basis of a volumetric digital medical image of the lung, composed of a voxel cloud. The method includes determining quantitatively and/or delineating a first volume A of the cloud in which spatial density of voxels is between −950 HU and −450 HU; and determining quantitatively and/or delineating a second volume B of the cloud in which spatial density of voxels is between −750 HU and −450 HU. A third volume C is determined quantitatively and/delineated, in addition or alternatively to volume B, defined as the differential volume (volume A−volume B); and the ratio (volume B/volume A) is computed and/or the ratio (volume C/volume A) is computed. The method includes comparing the value of these (this) ratio and/or the value of said volumes with one or several threshold values and exploiting the result of that comparison or these comparisons as said indicative parameter(s).
1. An automatic determination method of at least one parameter indicative of the functionality degree or level of a lung according to the ventilation rate of its cells, on the basis of a volumetric digital medical image of the lung, composed of a voxel cloud, said method comprising the steps of:
determining quantitatively and/or delineating a first volume A of the cloud in which spatial density of voxels is between approximately between −900 HU and −500 HU;
determining quantitatively and/or delineating a second volume B of the cloud in which spatial density of voxels is approximately between −700 HU and −500 HU;
determining quantitatively and/or delineating a third volume C corresponding to a volume in which the spatial density of voxels is approximately between −900 HU and −700 HU, and also defined as the differential volume [volume A−volume B];
computing the ratio [volume B/volume A] and/or computing the ratio [volume C/volume A];
comparing the value of these (this) ratio with one or several predetermined threshold values and exploiting the result of that comparison or these comparisons as said indicative parameter(s),
processing voxel cloud(s) by executing software operators configured to segment skin, trachea, volume A and volume B, preferably in that order, and
retrieving the partial volume effect and the potential isolated non-representative voxels for volumes B by realizing one or several mathematical operators, typically mathematical morphology operators such as an erosion followed by a dilation (defined as an opening) of one 1 voxel in 6-neighborhood.
2. The automatic determination method according to claim 1 , wherein said method is simultaneously or successively applied to voxel clouds of both lungs of a patient, to deduce an indicative parameter of the patient's respiratory situation.
3. The automatic determination method according to claim 1 , wherein said method further comprises quantitatively determining and/or delineating another volume in which the spatial density of voxels is between approximately −500 HU and approximately −100 HU.
4. The automatic determination method according to claim 1 , wherein said method further comprises quantitatively determining and/or delineating another volume in which the spatial density of voxels is between approximately −100 HU and approximately 100 HU.
5. The automatic determination method according to claim 1 , wherein said method further comprises quantitatively determining and/or delineating another volume in which the spatial density of voxels is between approximately −1000 HU and approximately −900 HU.
6. The automatic determination method according to claim 1 , wherein said method is realized, at least for image segmentation operations, by executing a neural network, in particular a convolutional neural network.
7. The automatic determination method according to claim 6 , wherein said convolutional neural network is U-Net.
8. The automatic determination method according to claim 1 , wherein said method uses one or several indicative parameters, automatically or not, in connection with one or several other existing parameters or criteria describing patients' health or physiology, in order to sort patients into a severity level of a given pathology affecting the functionality degree or level of the lungs, which will allow a physician to have more informative data to optimize a therapy proposal.
9. A computer or data processing system comprising means for carrying out the steps of the method of claim 1 , and configured to provide at least one indicative parameter.
10. A non-transitory computer readable medium comprising instructions which, when executed by a computer system, causes said system to carry out the steps of the method of claim 1 .