Methods and systems for classifying a medical image dataset

1 . A computer-implemented method for classifying a medical image dataset, the method comprising:

receiving the medical image dataset showing a body part of a patient;

providing a first classification stage configured to classify the medical image dataset as normal or not-normal;

providing a second classification stage different than the first classification stage configured to classify the medical image dataset as normal or not-normal;

subjecting the medical image dataset to the first classification stage to classify the medical image dataset as normal or not-normal; and

subjecting the medical image dataset to the second classification stage to classify the medical image dataset as normal or not-normal if the medical image dataset is classified as normal by the first classification stage, wherein

at least one of the first classification stage or the second classification stage includes segmenting at least one compartment from the medical image dataset to define a segmented dataset,

at least one of the first classification stage or the second classification stage is configured to independently classify i) image data of the medical image dataset as normal or not-normal and ii) image data of the segmented dataset as normal or not-normal, and

the medical image dataset is classified as normal if the image data of the medical image dataset is classified as normal and the image data of the segmented dataset is classified as normal.

2 . The method of claim 1 , wherein:

the first classification stage comprises inputting the medical image dataset to a first trained classifier configured to determine that the medical image dataset is normal, and

the second classification stage comprises inputting the medical image dataset to a second trained classifier different than the first trained classifier, the second trained classifier configured to recognize medical abnormalities in medical image datasets.

3 . The method of claim 2 , further comprising:

obtaining a prior medical image dataset, the prior medical image dataset showing the body part of the patient at a different point in time as the medical image dataset,

wherein at least one of the first classification stage or the second classification stage is configured to classify the medical image dataset as normal or not-normal further based on the prior medical image dataset.

4 . The method of claim 3 , further comprising:

obtaining at least one image quality parameter for the medical image dataset; and

classifying the medical image dataset as not-normal based on the at least one image quality parameter.

5 . The method of claim 4 , further comprising:

obtaining supplementary data associated with the medical image dataset, wherein

at least one of the first classification stage or the second classification stage are configured to classify the medical image dataset further based on the supplementary data.

6 . The method of claim 1 , wherein at least one of the first classification stage or the second classification stage comprises:

providing a plurality of different specifically trained classifiers each configured to classify image data of a respective segmented dataset as normal or not-normal for a respective compartment of the respective segmented dataset,

selecting one specifically trained classifier from the plurality of different specifically trained classifiers according to the at least one compartment of the segmented dataset, and

classifying the image data of the segmented dataset as normal or not-normal based by applying the one specifically trained classifier to the image data.

7 . The method of claim 1 , further comprising:

obtaining a prior medical image dataset, the prior medical image dataset showing the body part of the patient at a different point in time as the medical image dataset,

wherein at least one of the first classification stage or the second classification stage is configured to classify the medical image dataset as normal or not-normal further based on the prior medical image dataset.

8 . The method of claim 7 , wherein at least one of the first classification stage or the second classification stage is further configured to classify the medical image dataset as normal or not-normal based on determining a change between the medical image dataset and the prior medical image dataset.

9 . The method of claim 1 , further comprising:

obtaining at least one image quality parameter for the medical image dataset; and

classifying the medical image dataset as not-normal based on the at least one image quality parameter.

10 . The method of claim 1 , further comprising:

obtaining supplementary data associated with the medical image dataset, wherein

at least one of the first classification stage or the second classification stage are configured to classify the medical image dataset further based on the supplementary data.

11 . The method of claim 10 , further comprising:

adjusting a sensitivity of at least one of the first classification stage or the second classification stage based on the supplementary data.

12 . The method of claim 10 , further comprising:

checking if one or more compartments of relevance can be established based on the supplementary data;

segmenting one or more actual compartments from the medical image dataset;

checking if the one or more compartments of relevance are comprised in the one or more actual compartments; and

at least one of,

classifying the medical image dataset as not-normal if no compartment of relevance can be established, or

classifying the medical image dataset as not-normal if at least one of the one or more compartments of relevance is not comprised in the one or more actual compartments.

13 . The method of claim 10 , wherein the supplementary data includes at least one of demographic data of the patient, a diagnostic task to be performed for the patient, lab data of the patient, or a medical report of the patient.

14 . The method of claim 1 , wherein at least one of the first classification stage or the second classification stage comprises:

calculating a confidence value for a classification result of classifying the medical image dataset as normal or not-normal, and

classifying the medical image dataset as not-normal if the confidence value is below a predetermined threshold.

15 . The method of claim 1 , further comprising:

modifying a worklist of a user based on a classification result of classifying the medical image dataset as normal or not-normal, the worklist comprising a task for the user associated with medical image dataset.

16 . The method of claim 1 , further comprising:

generating a medical report based on a classification result of classifying the medical image dataset as normal or not-normal; and

providing the medical report.

17 . A non-transitory computer-readable medium on which program elements are stored that, when executed by a computing unit of a system for classifying a medical image dataset, cause the system to perform the method of claim 1 .

18 . A system for classifying a medical image dataset comprising:

an interface unit; and

a computing unit, wherein the computing unit is configured to cause the system to:

receive the medical image dataset showing a body part of a patient via the interface unit,

provide a first classification stage configured to classify the medical image dataset as normal or not-normal,

provide a second classification stage different than the first classification stage configured to classify the medical image dataset as normal or not-normal,

subject the medical image dataset to the first classification stage to classify the medical image dataset as normal or not-normal,

subject the medical image dataset to the second classification stage to classify the medical image dataset as normal or not-normal if the medical image dataset is classified as normal by the first classification stage, and

provide a classification result of the medical image dataset being normal or not-normal via the interface unit, wherein

at least one of the first classification stage or the second classification stage includes segmenting at least one compartment from the medical image dataset to define a segmented dataset,

at least one of the first classification stage or the second classification stage is configured to independently classify i) image data of the medical image dataset as normal or not-normal and ii) image data of the segmented dataset as normal or not-normal, and

the medical image dataset is classified as normal if the image data of the medical image dataset is classified as normal and the image data of the segmented dataset is classified as normal.

19 . A computer-implemented method for classifying a medical image dataset, the method comprising:

receiving the medical image dataset showing a body part of a patient;

providing a first classification stage configured to classify the medical image dataset as normal or not-normal;

providing a second classification stage different than the first classification stage configured to classify the medical image dataset as normal or not-normal;

subjecting the medical image dataset to the first classification stage to classify the medical image dataset as normal or not-normal;

subjecting the medical image dataset to the second classification stage to classify the medical image dataset as normal or not-normal if the medical image dataset is classified as normal by the first classification stage;

obtaining supplementary data associated with the medical image dataset, checking if one or more compartments of relevance can be established based on the supplementary data;

segmenting one or more actual compartments from the medical image dataset;

checking if the one or more compartments of relevance are comprised in the one or more actual compartments; and

at least one of,

classifying the medical image dataset as not-normal if no compartment of relevance can be established, or

classifying the medical image dataset as not-normal if at least one of the one or more compartments of relevance is not comprised in the one or more actual compartments.