1. U.S. Pat. No. 6,630,937 October 2003 Kallergi et al. “Workstation interface for use in digital mammography and associated method”
2. Robert A. Schmidt “Digital Mammography, networking, PACS, and Dante's Inferno” Applied Radiology, Supplement to September 2006, p21-25.
3. Gillian M. Newstead “Digital Mammography: Cost and Workflow Issues” Applied Radiology, Supplement to September 2006, p17-20.
4. Eric A. Berns et al. “Digital and Screen-Film Mammography: Comparison of Image Acquisition and Interpretation Times” AJR Women's Imaging: 187 July 2006, p38-41.
5. Laszlo Tabar et al “Teaching Atlas of Mammography”, Thieme Stuttgart, New York 2001
6. Laszlo Tabar et al “Breast Cancer: The Art and Science of Early Detection with Mammography—Perception, Interpretation, Histopathologic Correlation” 2005 Georg Thieme Verlag
Not Applicable.
Not Applicable
The present invention relates generally to the field of medical imaging systems. Particularly, the present invention relates to a method and apparatus for a soft-copy reading mammography interpretation workstation.
The U.S. patent Classification Definitions: 345/418 (class 345, Computer Graphics Processing and Selective Visual Display Systems, subclass 418 Computer Graphics Processing); 345/501 (class 345 Computer Graphics Processing and Selective Visual Display Systems, subclass 501 Computer Graphic Processing System); 378/37 (class 378, X-Ray or Gamma Ray System or Devices, subclass 37 Mammography); 382/173 (class 382 Image Analysis, subclass 173 Image Segmentation).
Digital mammography images are very data-intensive. A standard 4-view study acquired on a digital system with a 50 μm detector can contain more than 140 MB of data. When displaying such mammographic images on dual 5 MP (5 mega pixel) monitors, the radiologist is really seeing only 7% of the information. One of the issues that radiologists face is that many of the tools that they use, such as an interpretation workstation, are relatively primitive when it comes to dealing with such large amounts of data ([1]).
Screening mammography reading requires high throughput due to the large volume of screen mammograms and the low reimbursement rate. Historically, screening mammograms on film tend to be read in batches with the average interpretation time per case being around 1 minute, including dictation. However the median film reading time may be as short as 15 seconds because the vast majority of screening cases are negative.
A number of studies ([2] and [3]) have showed that radiologists take a significantly longer time for interpretation when reading digital mammography cases as compared to the time for screen-film mammography. Reasons for this increase in reading time include the poor performance of data transactions and inappropriate viewing workflows for mammography. These needs are not well- addressed in current PACS workstations.
This invention provides a method and apparatus to overcome the limitations from current PACS workstations by providing a system that
(1) customizes mammographic-specific reading and hanging protocol sequences for both screening and diagnostic study;
(2) streamlines reading workflow from opening a study to generating a report using a single button and mouse wheel;
(3) compares current exam with multiple prior exams and each prior exam is accessible by single click;
(4) high-lights the current images in all hanging protocols to minimize the risk of misdiagnosis from priors;
(5) incorporates expert film viewing methodology into digital, including Tabar's systematic viewing masks to enhance perception of subtle radiographic abnormalities;
(6) navigates by mouse wheel to view full-resolution image and visually traces pixels that have been viewed—no need to manually pan and zoom the images to view all pixels at full resolution;
(7) allows user to use electronic grease pen to markup the image any time; and automatically generates a corresponding mammography recall form and screening report;
(8) integrates the markup information with Reporting System.
This invention provides a simple user interface and sequencing viewing method for a mammography interpretation workstation. In particular, the invention includes the method and apparatus that moves the source images and the associated data using two-level of pre-fetching and caching mechanism, sequences the reading workflow (including worklist, hanging protocol, viewing sequencing), draws markup using electronic grease pan, and automatically generates the recall forms and diagnosis reports. The user interface operates on single button and mouse wheel style to maximize the radiologists' efficiency.
The workstation consists of two display screens: navigation screen and image screen (although two screens may physically reside on a same monitor; or one of two screens may physically reside on more than two monitors). The navigation screen (see
The reading sequence includes:
Step 1—Overall viewing of 8 standard 4-view mammograms from current and prior exams. This step provides an overall viewing for radiologists to make breast composition estimation (BIRADS breast density). Overall viewing of current and prior views also enhances the detection of tissue density changes; and overall viewing of the CC and MLO alternative views enforces the detection on both view projections.
Step 2—Assess image quality. The workstation provides bilateral or unilateral 2-view hanging protocols and a check list for quality assurance. The check items can be entered by users, which typically include checking if missing any images; if properly positioned with adequate breast compression; if sufficient image quality for make diagnosis; and if images are labeled correctly.
Step 3—Same view or breast comparison between current exam and prior exam to enhance the detection of tissue density changes and calcification appearance or disappearance on the same view.
Step 4—Systematic perception viewing, which is based on the expert viewing methodology that has been using for viewing films.
Step 5—Full-resolution all pixel image viewing for efficient searching through all pixels in an image.
Step —Finding review and report generation.
Each step has a number of associated hanging protocols, such as, 2×4 layout with current exam of 4 images display on left and prior exam of 4 images on right for over viewing step; or 1×2 layout with 1 current image on left and prior on right for same view comparison step etc.
User uses a visual configuration tool to define his or her viewing step in the viewing sequence, and the corresponding hanging protocols for each step (see
Each hanging protocol with each viewing step is stepped by a single “moving forward” button (or “backward” button). This button is available on both navigation screen and image screen (see
Any operation that might be used within each hanging protocol and viewing step can be performed by clicking the up or down button, or equivalently using the arrow key or mouse wheel. For example, in the step of systematic viewing, the mouse wheel moves the mask up or down. Or in the step of full-resolution all pixel viewing, the mouse wheel moves the trace of viewing window up or down.
The key design to ensure that image data is available on the screen almost immediately is to optimize the performance seen by the user once the user opens next case. In order to achieve this impression, the system performs so-called two-level on-demand pre-fetching/caching mechanisms:
The first level of the on-demand method ensures that the data resides on the workstation prior to the initiation of the display. This mechanism is referred to as “pre-fetch” of images. While it can be imagined that images can be retrieved rapidly on demand from an archive, existing systems, and certainly sites with high workloads on their networks, will not see usable interactive speed without pre-fetching. Thus the architecture designed here will explicitly include pre-fetching, utilizing a “reading worklist” to order the cases for retrieval.
The second level of the on-demand method ensures that the data is in the computer memory once the user navigates to a case by providing an explicit mechanism to make the next available case resident in system memory prior to the user navigating to that case. The mechanism to get a case into memory prior to its display is here referred to as “caching”.
This architecture is plotted in
The workstation is able to simultaneously display a set of current and prior screening mammogram images. In order to minimize the risk of misdiagnosis from prior exam, all the current images are high-lighted with a frame. This applies to all hanging protocols (see
The stepping button can be used to step through each prior exam. The system can also be configured so a separate looping button can be used to step through each prior.
The systematic viewing approach developed for the workstation is based on the expert viewing methodology that has been using for viewing films (see [4] and [5]). The essence of screening is to confidently reassure patients that their normal mammograms are indeed normal, while at the same time to find the occasional abnormality with great reliability. The systematic viewing method enables the radiologist to differentiate with confidence between the normal and the abnormal tissue by accounting for the radiopaque structure components that are seen against the radiolucent background.
The idea of the systematic viewing is to use two types of “mask” to enhance perception of subtle radiographic abnormalities. The horizontal mask is moved step-by-step, cranial-caudal either on the bilateral MLO views or the bilateral CC views. The oblique or vertical mask is moved away from the pectoral muscle or chestwall on the bilateral MLO views or CC views. By use of the masks, special attention can be given to asymmetric densities in regions with a high probability of malignant lesions (see
With the currently available monitors (5 MP), only a portion of an image can be displayed at one time at full resolution (in actual pixels). Manually panning and zooming on the computer monitor, while possible, is not natural for radiologists, it can be inefficient, and it is physically tiring and time-consuming. Automation of the viewing path for actual pixels will help to improve reading speed and reduce eye strain and finger tiredness.
The automation of the path to scan a full resolution image can be calculated from the image dimension and the display resolution in order to cover the full image in an electronically magnified viewport (see
The electronic magnify glass can be manually roam through an image under control of the user. The workstation then provides visual information if all actual pixels in the images has been roamed through by the electronic magnify glass (see
The workstation provides an “electronic grease pen” to allow the user to markup on the image display in any hanging protocol in any viewing step (see