Methods, systems and computer program products for classifying image data for future mining and training
A method for segmenting images is provided including tessellating an image obtained from one of an image database and an imaging system into a plurality of sectors; classifying each of the plurality of sectors by applying one or more pre-defined labels to each of the plurality of sectors, wherein the pre-defined labels indicate at least one of an image quality metric (IQM) and a metric of structure; assigning each of the plurality of classified sectors an Image Quality Classification (IQC); identifying anchor sectors among the plurality of classified sectors, applying filtering and edge detection to identify target boundaries; applying contouring across contiguous sectors and using the assigned IQC as a guide to complete segmentation of an edge between any two identified anchor sectors; and smoothing across segmented regions to increase parametric second-order continuity.
1 . A system for managing, visualizing and processing image data, the system including stored data including a master catalog of images each having a set of associated metadata and a module that organizes images into a plurality of collections and utilizes images belonging to the plurality of collections of images within a plurality of projects, wherein the system further comprises one or more relational databases with endpoints that cause the system to:
register images and associate metadata with the registered images in the master catalog;
register subsets of images registered to the master catalog to one or more of the plurality of collections; and
register operations on images that are registered to a collection to one or more of the plurality of projects registered to the collection.
2 . The system of claim 1 , wherein the set of metadata associated with the images in the master catalog includes information related to one or more of:
a subject of the image;
a modality of a device that acquired the image;
a manufacturer, brand, and/or model of the device that acquired the image;
a specific instance of the device that acquired the image;
an experiment, study, and/or purpose for acquiring the image;
a date on which the image was acquired; and
a time at which the image was acquired.
3 . The system of claim 1 , wherein the module further comprises a set of user interface modules for performing actions on the master catalog, the plurality of collections, the plurality of projects, and images and wherein the actions comprise one or more of:
filtering the master catalog using available metadata to curate sets of images and assigning the filtered image sets to one or more of the plurality of collections;
creating one or more projects for managing operations limited to images within one of the plurality of collections, wherein operations on images of a collection in a first project are independent of operations on images of a same collection in a second project;
visualizing images within a project using one or more of the user interface modules; and
applying annotations to images within a project using one or more of the user interface modules.
4 . The system of claim 3 , wherein the user interface module for applying annotations includes a module for defining a plurality of annotation tables, a module for associating an annotation table to an image, and a module for selecting annotation from the associated table and applying the selected annotation to the image.
5 . The system of claim 4 , wherein the scope of an annotation table is one of GLOBAL indicating that the annotation is available without restriction for use within a CATALOG or LOCAL to a COLLECTION indicating that the annotation is constrained to a given COLLECTION and all PROJECTS therein or LOCAL to a PROJECT only.
6 . The system of claim 4 , wherein annotation labels are qualitative and provide an indication of the quality of an image, the presence of a feature within an image, or the presence of pathophysiological state inferred from the image.
7 . The system of claim 4 , wherein annotation labels are quantitative and associate measurable information with an image or with one or more locations within an image.
8 . The system of claim 3 , wherein annotations are user-applied or algorithmically applied to an image.
9 . The system of claim 8 , wherein applied annotations are automatically registered to a database of the system and wherein automatically registering comprises associating the annotation to the Catalog, and to a Subject of the Image, and to metadata associated with the acquisition of the Image, including one or more of:
a modality of a device that acquired the image;
a manufacturer, brand, or model of the device that acquired the image;
a specific instance of the device that acquired the image;
an experiment, study, or purpose for acquiring the image;
a date at which the image was acquired; and
a time at which the image was acquired.
10 . The system of claim 3 , wherein applied annotations are automatically registered to a database of the system and wherein automatically registering comprises associating the annotation to the Image, the Project, and the Collection.
11 . The system of claim 3 , wherein a series of Projects associated with a Collection are used to:
record manual annotations by individual graders on a common Collection of Images;
record algorithmically-applied annotations on a common Collection of Images; and
allow individual graders to make modifications to annotations on a common Collection of Images; and
wherein a first Project may be duplicated to a second Project with annotations of the first Project copied to the second Project; and
wherein the annotations in the second Project are modified without impacting the annotations of the first Project.
12 . The system of claim 1 , wherein operations on images includes one or more of:
pre-processing images prior to visualization or subsequent analysis of the images,
rescaling images responsive to calibration information related to an image acquisition device, or structural information related to a subject of the image;
averaging images;
decimating images; and
exporting images from projects to a batch processing pipeline.
13 . The system of claim 1 , wherein registering a subset of images to one or more collections comprises allocating images for training, testing, and/or validating an algorithm, and wherein registering subsets includes one or more of:
assigning a first subset of images to a training Collection;
assigning a second subset of images to a testing Collection; and
assigning a third subset of images to be a validation Collection.
14 . The system of claim 13 , wherein allocating images to training sets comprises:
assigning a classification to each image according to a pre-defined classification metric; and
assigning images to separate Collections specific to each classification metric.
15 . A system for managing, visualizing and processing image data, the system including an image database; a collections database; a projects database; a methods Library; an outputs and reports database; and application programing interfaces (APIs) to communicate with processors external to the system, the system further comprising:
an image management module that organizes a catalog of images into a plurality of collections and projects assigned to collections;
an image visualization and annotation module that allows users to view images in context of the collections and projects assigned to collections;
an annotation module that enables a user to apply a plurality of qualitative and quantitative annotations to images,
an image processing module that applies a sequence of one or more algorithms or recipes, wherein a recipe includes a sequence of algorithm processes with specified parameters, to a filtered set of images within the collection or project; and
a data analysis module that includes parallel processing of image sets defined by a collection, with separate algorithms or recipes applied within a series of projects associated with a collection, for automated comparison with and between methods defined by the algorithms or recipes unique to each project associated with common data set defined by a collection.
16 . The system of claim 15 , wherein the system comprises further a systematic user interface, workflow, and data structure to facilitate a uniform record of image annotation and processing tasks to enable study of complex relationships within images and associated subject populations possible.
17 . The system of claim 15 , wherein the image management module further:
filters images obtained from a database or imaging system using a set of pre-defined attributes;
creates re-useable collections of the filtered images;
assigns images to projects; and
provides algorithmic and user applied labels and annotations to images within the assigned projects.
18 . The system of claim 15 , wherein the image management module further:
filters image according the annotations to provide sub-filter images; and
applies computational recipes to the sub-filtered images according to defined algorithms.
19 . The system of claim 15 , wherein the image visualization and annotation module automates decimation of images in sectors and presents sectors of images to the user for visualization and annotation.
20 . The system of claim 15 , wherein the image visualization and annotation module automates presentation of image pairs to a viewer for pairwise comparison and records a result of comparison.