Image reconstruction
A method of creating an image representative of a measured dataset by iteratively updating a base image includes: generating a principal dataset from the measured dataset, the principal dataset having noise at substantially the same level as the measured dataset; generating an additional dataset from the measured dataset such that the additional dataset has noise at substantially the same level as the measured dataset and is not identical to the principal dataset; processing the base image without noise compensation using the principal dataset and additional dataset to obtain a principal interim image and an additional interim image; comparing the principal and the additional interim image to determine an indication of a noise level present; and using the determined indication of noise present to select noise compensation to apply when processing the base image using the measured dataset to create a new base image representative of the measured dataset.
1. A computer implemented method of creating an image representative of a measured dataset by iteratively updating a base image, said method comprising:
using said measured dataset as a principal dataset or generating a principal dataset from said measured dataset by generating a bootstrapped dataset from said measured dataset;
generating at least one additional different dataset from said measured dataset, said additional different dataset comprising: a dataset generated from said measured dataset such that each additional different dataset has noise at substantially the same level as said measured dataset and such that each additional different dataset is not identical to the principal dataset by creating one or more bootstrapped datasets sampled from said measured dataset;
processing the base image without noise compensation using said principal dataset to obtain a principal interim image and processing the base image without noise compensation using each additional dataset to obtain at least one additional interim image;
comparing said principal and said at least one additional interim image to determine an indication of a level of noise present; and
using said determined indication of noise present to select noise compensation to apply when processing the base image using said measured dataset to create a new base image representative of said measured dataset;
wherein comparing said principal and at least one or more additional interim images to determine an indication of noise present comprises:
implementing an objective function according to which the goal is to find one or more parameters for a noise compensating procedure which, when found and applied to said one or more additional interim images, leads to an update of each additional interim image such that their collective difference from the principal interim image is minimized; and
wherein parameters of said noise compensation are modified to cause a greater level of noise compensation than that found by an optimization procedure of the objective function in order to slow down iterative updates of said base image and avoid accumulation of residual noise and in which continued iterative updates are such that said parameters of said noise compensation implemented are selected to be increasingly close to said parameters found by said optimization procedure of the objective function and that said parameters are selected to correspond to the greatest level of noise compensation found from all previous uses of said optimization procedure in the continued iterative updates.
2. A method of creating an image according to claim 1 , wherein said method comprises: using said selected noise compensation for producing an iterative update of said base image based on said measured dataset, in order to obtain said new base image representative of said measured dataset and optionally wherein said iterative update comprises an additive or multiplicative update.
3. A method of creating an image according to claim 2 , comprising: processing said base image using a different selected noise compensation procedure; and comparing a created image obtained using said selected noise compensation procedure and said different selected noise compensation procedure to select which noise compensation procedure to apply when processing said base image using said measured dataset to create a new base image representative of said measured dataset, and optionally wherein selecting which noise compensation procedure to apply comprises: comparing an interim image obtained using said noise compensation procedure to said principal interim image.
4. A method of creating an image according to claim 1 , wherein said objective function comprises: a measure related to distance between said one or more additional interim images and said principal interim image, and optionally wherein said measure related to distance comprises: one of: a sum of squares distance, a Kullback-Leibler measure of distance, any norm as a measure of distance, or another appropriate cross-likelihood measure correlated to distance.
5. A method of creating an image according claim 1 , wherein said processing with noise compensation corresponds to a regularised iterative update derived from any regularised iterative image reconstruction algorithm using said measured dataset.
6. A non-transitory computer program product operable, when executed on a computer, to perform the method of claim 1 .
7. Imaging apparatus configured to create an image representative of a measured dataset from a base image by iteratively updating a base image, said apparatus comprising:
dataset generation processing logic configured to use said measured dataset as a principal dataset or generate a principal dataset from said measured dataset by generating a bootstrapped dataset from said measured dataset, said dataset generation processing logic being further configured to generate at least one additional different dataset from said measured dataset, said additional different dataset comprising: a dataset generated from said measured dataset such that each additional different dataset has noise at substantially the same level as said measured dataset and such that each additional different dataset is not identical to the principal dataset by creating one or more bootstrapped datasets sampled from said measured dataset;
processing logic configured to process the base image without noise compensation using said principal dataset to obtain a principal interim image and to process the base image without noise compensation using each additional dataset to obtain a at least one additional interim image;
comparison processing logic configured to compare said principal interim image and at least one additional interim image to determine an indication of a level of noise present; and image creation processing logic configured to use said determined indication of noise to select a level of noise compensation to apply when processing the base image using said measured dataset to create a new base image representative of said measured dataset;
wherein parameters of said noise compensation are modified to cause a greater level of noise compensation than that found by an optimization procedure of the objective function in order to slow down iterative updates of said base image and avoid accumulation of residual noise and in which continued iterative updates are such that said parameters of said noise compensation implemented are selected to be increasingly close to said parameters found by said optimization procedure of the objective function and that said parameters are selected to correspond to the greatest level of noise compensation found from all previous uses of said optimization procedure in the continued iterative updates.