This patent specification relates to medical ultrasound imaging. More particularly, this patent specification relates to processing and/or display of breast ultrasound information for breast cancer screening and/or diagnosis purposes.
Volumetric ultrasound scanning of the breast can serve as a complementary modality for breast cancer screening as described, for example, in the commonly assigned US 2003/0007598A1 and US 2003/0212327A1, each of which is incorporated by reference herein. Whereas a conventional two-dimensional x-ray mammogram only detects a summation of the x-ray opacity of individual slices of breast tissue over the entire breast, ultrasound can separately detect the sonographic properties of individual slices of breast tissue, and therefore may allow detection of breast lesions where x-ray mammography alone fails. Another well-known shortcoming of x-ray mammography practice is found in the case of dense-breasted women, including patients with high content of fibroglandular tissues in their breasts. Because fibroglandular tissues have higher x-ray absorption than the surrounding fatty tissues, portions of breasts with high fibroglandular tissue content are not well penetrated by x-rays and thus the resulting mammograms contain reduced information in areas where fibroglandular tissues reside.
Still another shortcoming of x-ray mammography practice relates to difficulty in imaging near the chest wall. A substantial number of cancers are known to occur within 3 cm of the chest wall. In the craniocaudal (CC) and mediolateral oblique (MLO) views common to x-ray mammography acquisition protocols, the breasts are flattened between compression plates extending outward from the chest wall, and the x-rays are directed through the compression plates in a direction roughly parallel to the chest wall. With regard to breast tissues near the chest wall, it is physically difficult to pull these tissues outward to locations between the compression plates, making it difficult to properly image these tissues and more likely that cancers in these tissues will be missed in the resultant x-ray mammograms.
In addition to being useful as a complementary modality to x-ray mammography, ultrasound mammography could well become a sole breast cancer screening modality for at least some patient groups. For example, it is believed that preventive health care policy will progress toward the adoption of regular breast cancer screening procedures for increasingly younger women, e.g., women under the age of 40, and perhaps even under the age of 30 if there is a family history of cancer. Because younger women generally have denser breasts, the shortcomings of conventional two-dimensional x-ray mammography are expected to become especially apparent. Even further, because the dangers of x-ray radiation exposure are cumulative over a lifetime, ultrasound mammography could well become a sole breast cancer screening modality for women in these younger age groups. Other demographics indicating higher breast densities among certain groups, regions, or countries may also lead to the increased adoption of breast ultrasound as a sole or adjunctive screening modality for those groups, regions, or countries.
Once a thorough set of breast ultrasound scans is obtained, a challenge arises in the context of processing and displaying the breast ultrasound information to a clinician. In general, there is an inherent tension between (i) promoting high sensitivity/specificity in the screening and/or diagnosis process, and (ii) promoting efficient patient throughput to keep costs manageable. Thus, for example, while careful slice-by-slice scrutiny of the raw ultrasound scans by a well-trained radiologist would promote high sensitivity and specificity, the overall workflow efficiency of this method would be low, and therefore costs would be high, in view of the hundreds of individual raw ultrasound slices to be reviewed for each patient.
Accordingly, it would be desirable to provide an interactive user interface for viewing breast ultrasound information that can be effective for (i) adjunctive ultrasound mammography environments in which the ultrasound information complements x-ray mammogram information, and/or (ii) ultrasound-only mammography environments in which ultrasound is a sole screening modality.
It would be further desirable to provide processing and display of breast ultrasound information in a manner that promotes high specificity and sensitivity in the breast cancer screening and/or diagnosis process.
A system, method, and computer program product for processing and displaying breast ultrasound information is provided, comprising receiving a three-dimensional data volume of a sonographic property of a chestwardly compressed breast and processing the data volume to generate a plurality of two-dimensional coronal thick-slice images, each coronal thick-slice image being representative of the sonographic property within a thick-slice subvolume of the breast substantially parallel to a, coronal plane, wherein the thick-slice subvolume has a thickness selected for optimal viewing of terminal ductal lobular unit (TDLU) patterns in a display of the thick-slice images. In one particularly useful preferred embodiment, the thick-slice subvolumes are roughly 2 mm in thickness, and the thick-slice images are displayed in cine fashion to a viewer in a succession corresponding to progressive depths toward and/or away from the chest wall. It has been found that the cine display of the thick-slice images provides striking facilitation of the perception of abnormal breast structures, such as those associated with cancer, when the thick-slice subvolumes are associated with the dimensions of breast TDLUs in terms of thickness.
According to another preferred embodiment, projection-view thick-slice images are displayed that correspond to slab-like subvolumes parallel to a projection-view plane, the projection-view plane being other than a coronal plane. Examples include CC-projection thick-slice images and MLO-projection thick-slice images, and are useful even though the breast volume was not flattened along those planes when the ultrasound volume was acquired. Preferably, the projection-view thick-slice images are displayed adjacent to corresponding x-ray mammogram views. According to another preferred embodiment, the projection-view thick-slice images are stretched, morphed, or otherwise remapped to have to have a lateral shape that more closely corresponds to the corresponding x-ray mammogram view, the resultant remapped projection-view thick-slice image thereby being more easily compared thereto to that x-ray mammogram.
During or after the scanning process, the ultrasound scan data is provided across a computer network 113 to an ultrasound server 114 that processes and generates display information according to the functionalities described herein. The ultrasound server 114 may perform other HIS/RIS (hospital information system/radiology information system) activities such as archiving, scheduling, etc. It is to be appreciated that the processing of the ultrasound scan data may be performed by any of a variety of different computing devices coupled to the computer network 112 in various combinations without departing from the scope of the preferred embodiments. According to a preferred embodiment, a viewing workstation 122 is provided that provides a cine display 124 of a progression of coronal thick-slice images to a clinician 121, each coronal thick-slice image representing a sonographic property of the breast within a slab-like thick-slice subvolume thereof generally parallel to a coronal plane having a thickness that is optimal for viewing terminal ductal lobular unit (TDLU) structures in the breast, one particularly useful thickness being roughly 2 mm. Alternatively or in conjunction with the cine display 124, an array 126 of coronal thick-slice images 128 can be displayed for allowing simultaneous viewing of multiple coronal thick-slice images. In another preferred embodiment (not shown), the clinician is also provided with the ability to view individual planar ultrasound slices (along sagittal, axial, coronal, or other cut-planes through the three-dimensional breast volume), and the ability to view x-ray mammogram images or images from other modalities, as described further in US 2003/0007598A1 and/or US 2003/0212327A1, supra.
The size and shape of the individual lobules vary, but most lobules are between 0.5 and 1.0 mm in diameter. The numerous acini within each lobule can be discerned and the associated terminal duct is easily visualized using subgross, thick-section histology. According to Tabar 2005, the average size of the individual lobule and the mean number of acini within individual lobules in the breast of an adult, premenopausal, nonpregnant woman will change during the phases of the menstrual cycle, with examples of average sizes being 1.06 mm in the proliferative phase, 1.24 mm in the early secretory phase, 1.82 mm in the late secretory phase, and 1.29 mm in the premenstrual phase. There is a complex coexistence of proliferation and regression of the lobules during the menstrual cycle. Beginning a few days before ovulation in each menstrual cycle, clusters of cells bud from the smaller ducts, form lumens, and grow rapidly into fully developed lobules and clusters. If pregnancy does not supervene, the newly formed structures atrophy over several months. Throughout menstrual life, there is almost continuous production and loss of tissue. A single biopsy of the breast may show budding lobules and mature lobules in various stages of regression. In the event of pregnancy, lobule formation and proliferation is intensified and becomes widespread throughout the breast
As described in 2005/0171430A1, supra, a coronal thick-slice image comprises an integration of a plurality of individual ultrasound slices lying within a coronal slab-like subvolume, i.e., coronal thick-slice subvolume. Thus, for example, where a coronal slab-like subvolume is represented by a three-dimensional voxel array V(x,y,z) of scalar values, the corresponding coronal thick-slice image would be a two-dimensional pixel array PCOR(x,y) of scalar values. In one preferred embodiment, each pixel value PCOR(x,y) is simply computed as an arithmetic average along the corresponding voxel column at (x,y) having the voxel values V(x,y,z0), V(x,y,z1), V(x,y,z2), . . . , V(x,y,zN), where N is the number of individual ultrasound slices lying in the coronal slab-like subvolume. Other techniques for integrating the component ultrasound slices into the coronal thick-slice images PCOR(x,y) according to the preferred embodiments include arithmetic averaging, geometric averaging, reciprocal averaging, exponential averaging, and other averaging methods, in each case including both weighted and unweighted averaging techniques. Other suitable integration methods may be based on statistical properties of the population of component ultrasound slices at common locations, such as maximum value, minimum value, mean, variance, or other statistical algorithms.
There can be a combination of both art and science in the perception of tissue structures within the breast by the clinician. The preferred embodiments described herein have been found to be strikingly advantageous for clinician perception of patterns relevant to the detection of breast abnormalities, such as those caused by cancer, by cueing to the human's ability to perceive certain types of visual harmonies and, more importantly, disturbance of those harmonies, even among otherwise random arrangements. A yard full of flowers, for example, might have a random arrangement of flowers, but the gardener can often detect a disharmony indicating that an irrigation pipe is leaking, or a disharmony indicating that an animal has tromped through, and so forth. In a similar way, cueing the thicknesses of the ultrasonic thick-slice subvolumes to observe TDLU patterns and structures provides for an analogous detection of disharmony there among by clinicians familiarized with normal, harmonious patterns and structures. The use of coronal thick-slice images is particularly useful because the ducts of the breast, to which the TDLUs are connected, all lead to the nipple, so there is an expected progression of their locations in the coronal thick-slice images to assist in the detection of the harmonies/disharmonies among the TDLUs, linear densities, and connective tissues.
As described previously, it is preferable that the breast have been chestwardly compressed during the acquisition of the volumetric ultrasound scans. One particular advantage of such chestward compression is that spiculated lesions will tend to flatten out in a direction generally parallel to the coronal plane. This, in turn, causes the spiculation structures to be more readily visible in the coronal thick-slice images.
The preferred embodiments described thus far in the instant specification have dealt primarily with coronal thick-slice imaging for a chestwardly-compressed breast. According to another preferred embodiment, it has also been found useful to display thick-slice images corresponding to slab-like subvolumes parallel to other planes, such as the craniocaudal (CC) and/or mediolateral oblique (MLO) planes (which are standard to x-ray mammography), even though the breast volume was not flattened along those planes and was instead compressed in the chestward direction when the ultrasound volume was acquired. Such thick-slice images are referenced herein as CC-projection thick-slice images, MLO-projection thick-slice images, LAT-projection thick-slice images, and so on, the term “projection” in these instances being used to signify that the breast was compressed along a plane that is not “native” to that view, i.e., a plane that is not parallel to the plane against which the breast is traditionally compressed for that view. It has been found that, especially when viewed in conjunction with the coronal thick-slice images, the CC-projection thick-slice images, MLO-projection thick-slice images, and/or other projection thick-slice images further facilitate enhanced visualization of one or more breast cancer subtypes. As with the coronal thick-slice images, it has been found particularly advantageous for the thicknesses of the corresponding slab-like subvolumes to be related to TDLU size in a manner similar to the thicknesses of the coronal slab-like subvolumes, supra.
Further improved perception of the breast structures and anatomical abnormalities therein is provided by the proximal display of the CC-projection thick-slice images 5031 and the CC-view x-ray mammogram 524. Advantageously, the single volumetric ultrasound scan 401 used for the coronal thick-slice images also provides the basis for the CC-projection thick-slice images 503i, such that additional volumetric ultrasound scanning along additional planes of compression is not required. Although it is preferable that the breast ultrasound volume 401 and the CC-view x-ray mammogram 524 were acquired in a same clinical visit, the scope of the preferred embodiments is not so limited, and indeed a variety of temporal combinations of the displayed information is within the scope of the preferred embodiments. Also illustrated in
Preferably, the x-ray mammogram views (or digitized versions thereof if in film format) are provided as inputs to the remapping algorithms to facilitate the remapping, as illustrated in
Whereas many alterations and modifications of the present invention will no doubt become apparent to a person of ordinary skill in the art after having read the foregoing description, it is to be understood that the particular embodiments shown and described by way of illustration are in no way intended to be considered limiting. By way of example, although primarily described supra in the context of ultrasound imaging, it is to be appreciated that data from other full-field breast imaging modalities (e.g., MRI, CT, PET) can be advantageously processed and displayed according to one or more of the described preferred embodiments. By way of further example, it is to be appreciated that substantially parallel to a coronal plane is used herein to generally reflect the practical realities of situations such as head-on scanning of the breast, and that there may be some deviation from the plane of the chest wall. For example, for a particular patient having highly pendulous breasts it might be found most optimal to compress the breast at some small angle, such as 15 degrees, away from the plane of the chest wall. In this case, slab-like subvolumes that are taken parallel to the plane of compression would still be considered substantially parallel to the coronal plane.
By way of still further example, although it is particularly advantageous to incorporate chestward breast compression and coronal thick-slice viewing of the TDLU-size-related thick-slice images, it is to be appreciated that the scope of the preferred embodiments is not necessarily so limited. In other preferred embodiments, for example, the breast could be compressed along an MLO or CC plane and the thick-slice subvolumes parallel to the MLO or CC plane, respectively. In still other preferred embodiments, compression could be omitted.
By way of even further example, it is to be appreciated that the scope of the preferred embodiments is not necessarily limited to choosing thick-slice thicknesses that optimize TDLU viewing, but rather can include any type of selection and processing of the thick-slice subvolume data that results in thick-slice images from which TDLU patterns are emphasized or otherwise readily apparent. For example, in alternative preferred embodiments, the thick-slice subvolumes can be thicker, but are processed (e.g., using CAD or other intelligent processing or voxel weighting scheme) such that TDLUs are emphasized. In one example, a structural emphasis algorithm similar to that described in the commonly assigned U.S. Pat. No. 7,103,205, which is incorporated by reference herein, can be adapted for emphasizing the TDLU structures. Therefore, reference to the details of the preferred embodiments are not intended to limit their scope, which is limited only by the scope of the claims set forth below.
This application claims the benefit of U.S. Provisional Application No. 60/826,545, filed Sep. 22, 2006, and U.S. Provisional Application No. 60/827,052, filed Sep. 26, 2006, each of which is incorporated by reference herein. The subject matter of this provisional patent specification is related to the subject matter of U.S. Ser. No. 10/997,293 (Pub. No. US 2005/0171430A1), which is a continuation-in-part of U.S. Ser. No. 10/305,936 (Pub. No. US 2003/0212327A1), which is a continuation-in-part of U.S. Ser. No. 10/160,836 (Pub. No. US 2003/0007598A1), which claims the benefit, inter alia, of U.S. Ser. No. 60/326,715, filed Oct. 3, 2001, each of which is incorporated by reference herein.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US07/20538 | 9/21/2007 | WO | 00 | 11/24/2009 |
Number | Date | Country | |
---|---|---|---|
60826545 | Sep 2006 | US | |
60827052 | Sep 2006 | US |