# US 8,787,637 B2 · Method for developing an information prediction device, use thereof, and corresponding storage medium and apparatus > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 8,787,637 B2 - Document ID: US8787637B2 - Status: Granted - Kind: B2 - Application number: 13/256,002 - Filed: 2010-03-12 - Published: 2012-07-05 - Granted: 2014-07-22 - Earliest claimed: 2009-03-12 - Assignees: Commisariat A L'energie Atomique Et Aux Energies Alternatives; Assistance Publique-Hopitaux de Paris - Applicants: Edouard Duchesnay; Marie-Laure Paillere; Arnaud Cachia; Jean-Luc Martinot; Eric Artiges - Inventors: Edouard Duchesnay; Marie-Laure Paillere; Arnaud Cachia; Jean-Luc Martinot; Eric Artiges - Web page: https://patentgrep.com/patent/US8787637B2 ## CPC - G06T 7/60: Analysis of geometric attributes - G06F 18/2113: by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation - G06F 18/2115: by evaluating different subsets according to an optimisation criterion, e.g. class separability, forward selection or backward elimination - G06V 10/771: Feature selection, e.g. selecting representative features from a multi-dimensional feature space - G06T 2207/10088: Magnetic resonance imaging [MRI] - G06T 2207/10104: Positron emission tomography [PET] - G06T 2207/30016: Brain - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract A method for developing a prediction device for predicting a phenotype of a person from imaging data of the person is provided. The method includes determining imaging descriptors, the determining imaging descriptors including acquiring multidimensional images of people with an imaging apparatus and extracting multidimensional image elements from the acquired multidimensional images to serve as descriptors The method also includes classifying the predetermined descriptors on the basis of the capability thereof to predict the phenotype, selecting, from among the classified descriptors, a relevant number of the best-classified descriptors that is sufficient to predict the phenotype and constructing the prediction device from the selected descriptors. ## Claims ### Claim 1 A method for developing a prediction device for predicting a biological, clinical or demographic trait of a person from imaging data of the person, the method comprising: determining imaging descriptors, the determining the imaging descriptors including: acquiring multidimensional images of people with an imaging apparatus; and extracting multidimensional image elements from the acquired multidimensional images to serve as descriptors; classifying the predetermined descriptors on the basis of the capability thereof to predict the trait; selecting, from among the classified descriptors, a relevant number of the best-classified descriptors that is sufficient to predict the trait, the selecting the relevant number of best-classified descriptors including calibrating a selection function of the number of descriptors on randomly permutated data; and constructing the prediction device from the selected descriptors. ### Claim 2 (depends on 1) The method according to claim 1 wherein the selecting the relevant number of best-classified descriptors comprises: selecting a predictive function; applying, for any different numbers of best-classified descriptors, a learning procedure specific to the predictive function chosen so as to determine parameters of the predictive function so as to optimize the prediction of the trait; calculating, for the any different numbers of best-classified descriptors, a selection function for the number of descriptors from parameters of the determined predictive function; and selecting the number of best-classified descriptors that optimizes the selection function. ### Claim 3 (depends on 2) The method according to claim 2 wherein the computing the selection function is based on a combination of a term for adjusting to the data of the predictive function penalized by a term that depends on the complexity of the predictive function. ### Claim 4 (depends on 2) The method according to claim 2 wherein the selecting the predictive function is based on a generative approach or a discriminating approach. ### Claim 5 (depends on 1) The method according to claim 1 wherein the acquired multidimensional images are three-dimensional images and the image elements are voxels. ### Claim 6 (depends on 5) The method according to claim 5 wherein the three-dimensional images are acquired by at least one of positron emission tomography and nuclear magnetic resonance. ### Claim 7 (depends on 1) The method according to claim 1 wherein the determining the imaging descriptors further includes pre-processing the acquired multidimensional images. ### Claim 8 (depends on 1) The method according to claim 1 wherein the classifying the descriptors leads to a list of overlapping descriptor subsets. ### Claim 9 (depends on 8) The method according to claim 8 wherein the classifying the descriptors is done using univariable methods such as a T test, an F test or an ROC analysis. ### Claim 10 (depends on 1) A method for a prediction device developed according to the method of claim 1, comprising: predicting the response of a person to treatment by transcranial magnetic stimulation from multidimensional images of the person's brain. ### Claim 11 A non-transitory computer-readable storage medium comprising a code for developing a prediction device intended to predict a biological, clinical or demographic trait of a person from imaging data of the person, the code comprising: instructions to determine imaging descriptors, the instructions to determine imaging descriptions comprising: instructions to acquire multidimensional images of people with an imaging apparatus; and instructions to extract multidimensional image elements from the acquired multidimensional images to serve as descriptors; instructions to classify the predetermined descriptors on the basis of the capability thereof to predict the trait; instructions to select, from among the classified descriptors, a relevant number of the best-classified descriptors that is sufficient to predict the trait, the instruction comprising an instruction to calibrate a function for selecting the number of descriptors on randomly permutated data; and instructions to construct the prediction device from the selected descriptors. ### Claim 12 An apparatus for developing a prediction device intended to predict a biological, clinical or demographic trait of a person from imaging data of that person, characterized in that it comprises: an imaging apparatus; and a data processing system comprising: a determiner for determining imaging descriptors, the means comprising: acquirer for acquiring multidimensional images of people by means of the imaging apparatus; and extractor for extracting multidimensional image elements from the acquired multidimensional images to serve as descriptors; classifier for classifying the predetermined descriptors on the basis of the capability thereof to predict the trait; selector for selecting, from among the classified descriptors, a relevant number of the best-classified descriptors that is sufficient to predict the trait, the means comprising means for calibrating a selection function for the number of descriptors on randomly permutated data; and constructor for constructing the prediction device from the selected descriptors.