Search

CN-115861168-B - Automated analysis of 2D medical image data with additional objects

CN115861168BCN 115861168 BCN115861168 BCN 115861168BCN-115861168-B

Abstract

The invention relates to automatic analysis of 2D medical image data with additional objects. A method for automatically analyzing 2D Medical Image Data (MID) comprising an additional object (EO) is described. According to the method, medical Image Data (MID) comprising an additional object (EO) is acquired from an examination part (ROI) of a patient (P) by a first modality (2), and Additional Image Data (AID) is acquired from the examination part (ROI) using a different modality (C). An automatic image analysis applicable to the additional object (EO) is performed based on the acquired Medical Image Data (MID) and the acquired additional image data (EID). Furthermore, an analysis device (40, 50) is described. A medical imaging system (1) is also described.

Inventors

  • Christian Schumer
  • Sven Martin Sutter
  • Celesh Condetti

Assignees

  • 西门子医疗有限公司

Dates

Publication Date
20260505
Application Date
20220926
Priority Date
20210924

Claims (10)

  1. 1. A method for automatically analyzing 2D Medical Image Data (MID) comprising an additional object (EO), Wherein the additional object comprises an External Object (EO), The method comprises the following steps: acquiring 2D Medical Image Data (MID) from an examination portion (ROI) of a patient (P) using a first modality (2), Acquiring additional image data (EID) from the examination portion (ROI) using a different modality (C), wherein the additional image data (EID) comprises external image data, Performing an automatic image analysis applicable to the additional object (EO) based on the acquired 2D Medical Image Data (MID) and the acquired additional image data (EID), Wherein the automatic image analysis comprises: -a separate first analysis of the additional image data (EID), and A subsequent second analysis of the Medical Image Data (MID) based on the result of the first analysis, It is characterized in that the method comprises the steps of, -The first analysis comprises: extracting information of the type and/or position of the External Object (EO) from the External Image Data (EID), Identifying a type (T) of the External Object (EO) and/or locating the External Object (EO) in the External Image Data (EID), -Selecting a dedicated Medical Image Analysis Application (MIAA) for identifying and/or locating an External Object (EO) in the Medical Image Data (MID) based on previous identification and/or localization, and -The second analysis comprises: -identifying and/or locating an External Object (EO) in the Medical Image Data (MID) based on the Medical Image Analysis Application (MIAA).
  2. 2. The method according to claim 1, wherein the first analysis comprises the step of identifying and/or locating an additional object (EO) in the additional image data (EID).
  3. 3. The method of claim 1, wherein, -The extracting step comprises detecting a route (CS) of an External Object (EO) in said External Image Data (EID), and -The step of selecting a dedicated Medical Image Analysis Application (MIAA) comprises adjusting the analysis application focused on the detected route (CS).
  4. 4. A method according to claim 3, wherein adjusting the analysis application comprises taking into account the route (CS) of the External Object (EO) by at least one of the following steps: -suppressing medical image areas of overlapping External Objects (EO), and/or Weighting the segmentation results or heat map regression results, -Processing the medical image subregions of the External Object (EO) with said overlap separately.
  5. 5. The method according to any of the preceding claims, wherein the first analysis comprises detecting the Insertion Position (IP) and/or the type (T) of the part of the extracorporeal device.
  6. 6. The method of claim 5, wherein, The partial extracorporeal device comprises a port catheter having a covering comprising a Pattern (PT) that varies slightly from tip to port device, -The first analysis comprises at least one of: Identifying the type (T) of the port catheter based on an overlay Pattern (PT), Positioning an Insertion Position (IP) of the port catheter based on the overlay Pattern (PT), -Estimating an Insertion Depth (ID) based on the overlay Pattern (PT).
  7. 7. An analysis device (40, 50), comprising: -a first input interface (41), the first input interface (41) being for acquiring 2D Medical Image Data (MID) comprising an additional object (EO) from an examination portion (ROI) of a patient (P) acquired by a first modality (2), wherein the additional object comprises an External Object (EO), -A second input interface (42), the second input interface (42) being for acquiring additional image data (EID) from the examination portion (ROI) using a different modality (C), wherein the additional image data (EID) comprises external image data, An analysis unit (43, 53), the analysis unit (43, 53) being adapted to perform an automatic image analysis adapted to the additional object (EO) based on the acquired 2D Medical Image Data (MID) and the acquired additional image data (EID), Wherein the automatic image analysis comprises: -a separate first analysis of the additional image data (EID), and A subsequent second analysis of the Medical Image Data (MID) based on the result of the first analysis, It is characterized in that the method comprises the steps of, -The first analysis comprises: extracting information of the type and/or position of the External Object (EO) from the External Image Data (EID), Identifying a type (T) of the External Object (EO) and/or locating the External Object (EO) in the External Image Data (EID), -Selecting a dedicated Medical Image Analysis Application (MIAA) for identifying and/or locating an External Object (EO) in the Medical Image Data (MID) based on previous identification and/or localization, and -The second analysis comprises: -identifying and/or locating an External Object (EO) in the Medical Image Data (MID) based on the Medical Image Analysis Application (MIAA).
  8. 8. A medical imaging system (1), comprising: a scanning unit (2), the scanning unit (2) being adapted to acquire Measurement Data (MD) from an examination portion (ROI) of a patient (P), A reconstruction unit (4 a), the reconstruction unit (4 a) being adapted to reconstruct image data (MID) based on the acquired Measurement Data (MD), An analysis device (40) according to claim 7, -An additional modality (C) for acquiring additional image data (EID) from the examination part (ROI).
  9. 9. A computer program product with a computer program which can be directly loadable into a memory means of a medical imaging system (1), the computer program having program segments for performing all the steps of the method according to any of claims 1 to 6 when the computer program is executed in the medical imaging system (1).
  10. 10. A computer readable medium having stored thereon program segments readable by and executable by a computer unit for performing all the steps of the method according to any of claims 1 to 6 when the program segments are executed by the computer unit.

Description

Automated analysis of 2D medical image data with additional objects Technical Field The invention relates to a method for automatically analyzing 2D medical image data comprising an additional object. The invention also relates to an analysis device. Furthermore, the invention relates to a medical imaging system. Background The placement and insertion of external devices is often used in clinical practice for life support purposes as well as patient monitoring. For example, endotracheal tubes are commonly used for ventilation, while intravenous tubes are commonly inserted for medication or pressure monitoring. One major challenge in analyzing two-dimensional medical images using external devices is projecting the extracorporeal and inserted objects into one image plane, which superimposes the density from the external device on the internal tissue structure. This effect complicates image interpretation by obscuring important findings by overlaying medical devices. It may be feasible to use deep learning to detect and identify devices, assuming that a sufficiently large training data set is available for the purpose of the image analysis application to learn essentially to distinguish between in vitro objects and inserted devices. A prominent example is the assessment of the position of a central venous catheter by researchers in Subramannian et al ,"Automated Detection and Type Classification of Central Venus Catheters in Chest X-rays",https://arxiv.org/abs/1907.01656; and Hansen et al ,"Radiographic Assessment of CVC Malpositioning:How can AI best support clinicians?",https://openreview.net/pdfid＝ImcP8kkqtfZ. However, such a strategy automatically results in the need to capture any type of variable generated by the extracorporeal device with a large training data set, which appears to be particularly challenging in the case of intensive care units, for example, with high intrinsic image diversity. A second possibility to improve the image analysis application is to segment any type of external object and the inserted object. For example, the authors in Lee et al ,"A Deep-Learning System for Fully-Automated Peripherally Inserted Central Catheter(PICC)Tip Detection",https://dx.doi.org/10.1007/s10278-017-0025-z. also identified the tip location of peripherally inserted central catheter by simultaneously segmenting and classifying other inserted devices and extracorporeal devices, at the cost of very high annotation complexity. A further possibility to consider various in vitro devices without using large training data sets and annotation work is to manually create synthetic cases. The risk of this approach by Yi et al ,"Automatic Catheter and Tube Detection in Pediatric X-ray Images Using a Scale-Recurrent Network and Synthetic Data",https://arxiv.org/pdf/1806.00921.pdf. is a potential mismatch between training conditions and test conditions due to conceptual drift. Therefore, there is a problem in that high quality is achieved when analyzing medical image data from an examination object including an extracorporeal device and an intracorporal device. Disclosure of Invention The aforementioned problems are solved by a method for automatically analyzing 2D medical image data comprising an additional object according to the inventive solution, by an analysis device according to the inventive solution and by a medical imaging system according to the inventive solution. According to a method for automatically analyzing 2D medical image data comprising an additional object, 2D medical image data is acquired from an examination portion (also referred to as region of interest) of a patient. In general, an additional object is defined as an object that is intended to be distinguished, but which cannot be distinguished or is difficult to distinguish from a part of the patient's body or other parts of the patient's body located at or near the same location as the additional object in the 2D medical image data based on information of the acquired 2D medical image data only. For example, the additional object comprises a foreign body, which is typically not part of the body of a healthy person or which comprises a material different from the tissue of a healthy person. In this case it must be explicitly pointed out that the expression "additional object" also includes the case of more than one additional object. In particular, the expression "additional object" includes various external objects, i.e. objects located outside the patient's body. For example, in a typical case, there may be a plurality of external objects on the body of the patient. Such external objects may include catheter ports, ECG devices (ecg=electrocardiogram), cables, endotracheal tubes or hoses, etc. The patient may be a human or an animal. The medical image may comprise, for example, an X-ray image, but also a 2D projection of a CT image or an MR image. However, different modalities are also used to acquire additional image data from the examinatio