User interface for X-ray tube-detector alignment
System (SYS) for supporting X-ray imaging and related methods. The system (SYS) comprises a machine learning module (MLM), a logic (LG) configured to compute output correction information for adjusting an imaging geometry of an X-ray imaging apparatus to achieve a target imaging geometry. A modulator (MOD,L-MOD, H-MOD, S-MOD) is the system is configured to provide a user instruction for imaging geometry adjustment. The user instruction is modulated based on the output correction information. The machine learning module was previously trained on training data including a specific user's responses to previous instructions.
1. A system for controlling X-ray imaging, comprising:
a memory for storing a plurality of instructions comprising a pre-trained machine learning model; and
an X-ray imaging apparatus comprising an X-ray source, an X-ray detector with no rigid mechanical coupling between the X-ray source and the X-ray detector, and controller circuitry;
wherein a processor is coupled to the memory and configured to execute the pre-trained machine learning model to compute output correction information for adjusting an imaging geometry of the X-ray imaging apparatus to achieve a target imaging geometry, wherein the controller circuitry is configured to provide a user instruction for the imaging geometry adjustment, modulated based on the output correction information, and wherein the processor is configured to execute the pre-trained machine learning model previously trained on training data including specific user's responses to previous user instructions for imaging geometry adjustments, and wherein multiple pre-trained or trainable machine learning models are associated with different users.
2. The system according to claim 1 , further comprising a user identification functionality configured to identify a user based on a credential and to cause selection of the machine learning model from the multiple pre-trained or trainable machine learning models based on the credential.
3. The system according to claim 1 , wherein:
the controller circuitry is configured to control a visible light projector to project a light beam onto a surface, the light beam modulated in accordance with the output correction information, and/or
the controller circuitry is configured to control a haptic vibrator to impart on a) a manual actuator mechanically coupled to the X-ray source of the X-ray imaging apparatus or b) on an electrical input communicatively coupled to the X-ray source of the X-ray imaging apparatus, mechanical vibrations being in accordance with the output correction information, and/or
the controller circuitry is configured to control an electroacoustic transducer to sound out the user instruction in accordance with the output correction information.
4. The system according to claim 1 , wherein, upon a change of a current imaging geometry, the user instruction is updated.
5. The system according to claim 1 , comprising a navigation unit configured to compute input correction information based on which the output correction information is computed by the machine learning model, the input correction information computed based on one of:
i) sensor readings provided by sensors arranged at the X-ray imaging apparatus;
ii) a scout image acquired using a collimator setting to effect a projection being detected at the X-ray detector of the X-ray imaging apparatus, the projection having a shape, the input correction information being computed based on a distortion of the shape relative to a pre-determined reference shape; and
iii) an intensity profile as detected at the X-ray detector and caused by an anti-scatter grid.
6. The system according to claim 3 , wherein modulation of light by the controller circuitry includes at least one of: a change of a light pattern cast by the light beam, a change in light gradient, and a change in light color and/or hue.
7. The system according to claim 3 , wherein the surface is in an examination region of the X-ray imaging apparatus.
8. The system according to claim 7 , wherein the surface is defined by a patient when present in the examination region.
9. A method for controlling X-ray imaging, comprising:
computing, by a processor configured to execute a pre-trained machine learning model, output correction information for adjusting an imaging geometry of an X-ray imaging apparatus, wherein the X-ray imaging apparatus is of the mobile type comprising an X-ray source and an X-ray detector with no rigid mechanical coupling between the X-ray source and the X-ray detector, to achieve a target imaging geometry, and
providing, by controller circuitry, a user instruction for imaging geometry adjustment, modulated based on the output correction information,
wherein the processor is configured to execute the pre-trained machine learning model is previously trained on training data including specific user's responses to previous user instructions for imaging geometry adjustments, and wherein multiple pre-trained or trainable machine learning models are associated with different users.
10. A non-transitory computer readable medium having stored thereon executable instructions which, when executed by at least one processor, cause the at least one processor to perform a method for controlling X-ray imaging, the method comprising:
computing, by a processor configured to execute a pre-trained machine learning model, output correction information for adjusting an imaging geometry of an X-ray imaging apparatus, wherein the X-ray imaging apparatus is of the mobile type comprising an X-ray source and an X-ray detector with no rigid mechanical coupling between the X-ray source and the X-ray detector, to achieve a target imaging geometry, and
providing, by controller circuitry, a user instruction for imaging geometry adjustment, modulated based on the output correction information, wherein the processor is configured to execute the pre-trained machine learning model previously trained on training data including specific user's responses to previous user instructions for imaging geometry adjustments, and wherein multiple pre-trained or trainable machine learning models are associated with different users.