Method and apparatus for three dimensional reconstruction of a joint using ultrasound

Inventors

Mahfouz, Mohamed R.

Assignees

JointVue LLC

Abstract

A method of generating a 3-D patient-specific bone model, the method comprising: (a) acquiring a plurality of raw radiofrequency (“RF”) signals from an A-mode ultrasound scan of a patient's bone at a plurality of locations using an ultrasound probe that comprises a transducer array; (b) tracking the acquiring of the plurality of raw RF signals in 3-D space and generating corresponding tracking data; (c) transforming each of the plurality of raw RF signals into an envelope comprising a plurality of peaks by applying an envelope detection algorithm to each of the plurality of raw RF signals, each peak corresponding with a tissue interface echo; (d) identifying a bone echo from the tissue interface echoes of each of the plurality of raw RF signals to comprise a plurality of bone echoes by selecting the last peak having a normalized envelope amplitude above a preset threshold, wherein the envelope amplitude is normalized with respect to a maximum peak existing in the envelope; (e) determining a 2-D bone contour from the plurality of bone echoes corresponding to each location of the ultrasound probe to comprise 2-D bone contours; (f) transforming the 2-D bone contours into an integrated 3-D point cloud using the tracking data; and, (g) deforming a non-patient specific 3-D bone model corresponding to the patient's bone in correspondence with the integrated 3-D point cloud to generate a 3-D patient-specific bone model.

Core Innovation

The invention generates a 3-D patient-specific bone model from a plurality of raw signals acquired from an A-mode ultrasound scan of a patient's bone at a plurality of locations using an ultrasound probe that comprises a transducer array. The acquiring of the plurality of raw signals is tracked in 3-D space using a position tracking device and corresponding tracking data are generated. Model-based signal processing is applied to each raw signal to transform processed signals into an envelope having a plurality of peaks, where each peak corresponds with a tissue interface echo.

Bone echoes are identified from the tissue interface echoes by selecting the last peak having a normalized envelope amplitude above a preset threshold, where the envelope amplitude is normalized with respect to a maximum peak existing in the envelope. A 2-D bone contour is determined from the plurality of bone echoes corresponding to each location of the ultrasound probe to comprise 2-D bone contours. The 2-D bone contours are transformed into an integrated 3-D point cloud using the tracking data.

A non-patient specific 3-D bone model corresponding to the patient's bone is deformed in correspondence with the integrated 3-D point cloud to generate the 3-D patient-specific bone model. In the described system, the non-patient specific bone model is a statistical atlas model using PCA eigenbones, and the deformation corresponds to fitting in eigenspace. The document further outlines analogous cartilage reconstruction using probabilistic processing and an SVM trained with MRI/ultrasound registration.

Claims Coverage

The document provides one independent claim directed to generating a 3-D patient-specific bone model from tracked A-mode ultrasound signals, including envelope-peak identification of tissue interface echoes and deformation of a non-patient specific 3-D bone model based on an integrated 3-D point cloud.

Overall, the claim coverage centers on tracked multi-location A-mode ultrasound signal acquisition, normalized last-peak tissue-interface/bone-echo selection, derivation of 2-D bone contours and integration into a 3-D point cloud, and deformation of a non-patient specific 3-D bone model to generate a 3-D patient-specific bone model.

Stated Advantages

Documented Applications