Breast Imaging - Science topic

Lack of training data could be another issue plus to what mentioned above

I would recommend going through this paper to understand the issue and the solution.

Dear James O'Callaghan,

Thank you very much for taking the initiative of building software and medical imaging repositories. I have already visited and created an account in #imagingQA. Indeed, this website will help the scientific community to have free access to medical images and discuss problems with researchers to find a solution.

Sincerely,

Md Mamunur Rahaman

Dear Sheeja,

You can use our database:

In case you decide to use it, please follow the license instructions and cite the works we recommend there.

Regards.

Dear William,

Thank you for your response.

Actually the light source that we simulate is a point source light that has spectral width of 50 nm with peak around 400 nm and the detector is photo multiplier tube.

Bests

To refine your image, try finding the centroids of the dark regions of a binary image bw (from a grayscale image img) as well as pixel values and bounding boxes, using

s = regionprops(bw, img, {'Centroid','PixelValues','BoundingBox'});

Here is a complete example:

img = rgb2gray(imread('peppers.png'));

bw = im2bw(img);

%% find centroids, region pixel values and bounding boxes:

s = regionprops(bw,img, {'Centroid','PixelValues','BoundingBox'});

figure, imshow(img),hold on;

numObj = numel(s);

for k = 1 : numObj

s(k).StandardDeviation = std(double(s(k).PixelValues));

text(s(k).Centroid(1),s(k).Centroid(2), ...

sprintf('%2.1f', s(k).StandardDeviation), ...

'EdgeColor','b','Color','r');

end

The attached image shows sample results.

Hi,

Have a look at

Current Medical Imaging Review

International Journal on Imaging Systems and Technologies

Probably the most reliable and clinically most relevant way of verifying registration results is to ask experts to identify the same landmark points in both the fixed and moving image and then compute the distance between the fixed and transformed moving points.

Similarly, you can segment relevant structures in both the fixed and moving image and compare the fixed segments with transformed moving segments. Hausdorff distance (mean, 95th percentile) is a good metric, tells you how well the surfaces of the segments are matched. Some people use Dice Similarity Coefficient for segment comparison, which is a very poor metric, as its value is highly dependent on shape of segments (but it's very easy to compute, probably that's why many people still report that in papers instead of Hausdorff distance).

Both point and segment based evaluation can be performed in 3D Slicer (http://www.slicer.org - free, open-source medical image visualization and analysis software).

For segment-based evaluation, install SlicerRT extension and use Segment Comparison module (it can compute both Hausdorff and Dice metrics).

For point-based evaluation, you create two markup fiducial lists, apply the computed transform to the moving list, harden the transform, and compute distances in Python - or save transformed point positions to file and compute distances in Excel, R, etc.

If you have any questions, post it to http://discourse.slicer.org, you typically get expert help within a few hours.

Thank you for your quick reply. I Know mias dataset, but most of images are scans of original images. I need the original images because not noisy affected. 

Greetings 

Hello,

This is such a good question in case of propagation in multilayer media. Before measuring the values, first diagnose the epsilon and mue for all materials you have, because they will create surface wave or surface plasmonic wave between two layer of different materials. Now, electromagnetic wave has two components of normal and transverse in case of electric and magnetic fields, so that we have two components of phase propagation constant. now, one way to calculate it, by measuring the phase velocity or dispersion characteristics of wave and other way first calculate gamma and then by use of this calculate beta.

I am sending one paper, which might be helpful for you.

Thanks,

Dear Ms. Jasna Davidovic,

Our hospital uses breast mammography / tomosynthesis for diagnosis of breast cancer in conjunction with USG. PET is indicated in inoperable cases of stage III & IV. At present we do not have PEM, hence we do not use it, instead if clearly indicated, we go ahead with PET.

PEM is "NOT" used for screening of breast cancer for the reasons already stated by me. The institutions having PEM also uses only for specific indications.

Regarding radiation burden.. It is much more hyped that it really is. A few simple measures can significantly reduce the radiation burden. Newer PET scanners with time of flight (TOF) techniques require half the dose of F-18 FDG for injection without any compromise in image quality and interpretation. All books refer to radiation dose given by non-TOF systems which is not correct. Secondly reducing the kV and mA of CT significantly reduces the dose depending on the weight of patient instead of using a blanket standard for everyone.

Regarding CESM.. the more we complicate the screening test, more it will move out of vogue. Screening test should be affordable, as simple as possible, readily available and can be done in primary settings without much support..

Sorry, but this question is way too broad to solicit any meaningful answer. There are multiple levels you can answer it on: do you want to judge for the whole mammogram or identify the locations of abnormalities? Do you want to focus on microcalcifications, on masses, on architectural distortions, on ...???

The field is old, there is a biannual conference IWDM that was in 20 years dedicated to the sole task of image generation and image processing in (digital) mammography, and for anyone wishing to know where the field is at: This is the history you have to cope with.

Please don't just go and throw some feature extraction plus some classifier on the problem, since with an almost 100% likelyhood you'll at best reproduce results that are out there since ages.

Greetings!

Yes of course. You must use for all method included MRI, US, MG

Please go through this article; http://www.ncbi.nlm.nih.gov/pubmed/11996066 

Also you may use plantago ovata husk gel to get any shape but certainly it needs some technique to form phantoms using this material. 

Best of luck.

Charles:

I have recently reviewed this particular issue, so I will present below a summary of my main findings:

Definitions: First some working definitions. A unifocal tumor type is defined when only one malignant focus is found in the breast, in contrast both to a multifocal type defined when more than one malignant focus occurs in the same quadrant, and to the multicentric type defined when more than one malignant focus is observed in more than one single quadrant.

Now I would argue that it is important to first note that within phenotypically undifferentiated general breast cancer populations, it was found in a large retrospective analysis of the GeparTrio, GeparQuattro, and GeparQuinto trials¹ that patients with multicentric tumors but not the patients with multifocal tumors had worse DFS and OS than the patients with unifocal tumors, although which outcomes were not inferior for the patients with multicentric or multifocal tumors if pCR was achieved or breast conservation was performed after neoadjuvant chemotherapy. And the BRENDA Study Group in another large retrospective multicenter cohort study² found that RFS was significantly worse for MF/MC BC patients compared to patients with unifocal tumors, while overall survival was significantly worse for multicentric patients but not for multifocal patients compared to patients with unifocal tumors, and more specifically there was in fact a strong association between improved RFS and OS in patients with multifocal breast cancer.

Note therefore that: (1) both these robust analyses, and numerous other earlier studies, confirmed that in breast cancer populations not subgrouped by molecular type (presence or absence of hormone-positivity, HER2-positivity, and TNBC or basal positivity), only multicentricity, not multifocality is associated with compromised prognosis and outcome; and (2) that multifocality may in fact confer improved outcome.

So the core question is, do these results also hold for the specialized TNBC or basal phenotype? The answer is strikingly in the negative, and I argue below that three robust studies, if not yet dispositive, provide exceptionally strong data that in the unique context of triple negative (TNBC) or basal type breast cancer, and unlike in the general undifferentiated BC populations, both multicentricity and multifocality confer increase risk of lymph node metastasis, and of death and compromised (shorter) survival outcomes.

Thus, the subtype-specific prognostic importance of multifocal or multicentric tumors (MF/MC) was recently confirmed in the Korean Breast Cancer Society study³ which found that there was a significant role of multifocal or multicentric tumor (MF/MC) in developing distant metastasis and death when the tumors were triple-negative (TNBC) subtype, with an increased risk of having lymph node metastasis and a significantly shorter overall survival (32% increased risk of death) when compared with patients with single invasive tumor (and note that the prognostic significance of MF/MC tumors in terms of metastasis-free survival was only seen in patients with triple-negative tumors).

This is further confirmed in recent joint Swedish/Taiwanese data⁴ which has demonstrated that multifocal distribution of the invasive component of breast carcinomas is associated with a 2.75-fold increased risk of cancer-related death compared with unifocal tumors, irrespective of molecular phenotype, with an increased risk of dying of the disease being 2.18-fold in HER2-positive, 2.33-fold in TNBC, and 4.07-fold in basal-like cancers, compared with patients who had luminal A carcinomas⁵.

For triple negative (TNBC) or basal type breast cancer:

(a) an increase risk of lymph node metastasis, and

(b) of death and compromised (shorter) survival outcomes, and with:

I would conclude therefore based on the above evidence that this makes it imperative that multicentric or multifocal disease in TNBC or basal breast cancer subpopulations be both monitored closely, and be treated aggressively to minimize the risk of compromised survival outcome.

'Size or locularity'  as  you  asked  above  appears  to  be  a  researcher  bias  to  me.

The  researcher  should  set  out  to find  out  what  factors (not  size  or  locularity) contribute  to  failed  per-cutaneous  drainage  of  breast  abscesses.

You  might  be  surprised  to  discover  a  factor  such  as  poor  technique.

In  my  opinion,   even  if  you  find  out  the  factor  that  has  a  higher   rate it  cannot  be  extrapolated  for  every  centre  or  environment  until  you  have  properly  tackled  the  issue  of   technique  which  might  crop  in  as  a  confounding  variable.

Thank  you.

I have the database for digital mammogram which is older one and the ground truth information is available

Thanks for the answers and the references. The purpose is to research optimal spectral CT of breast in terms of  scanning parameters, energy bin selection, reconstruction, and material decomposition. Intended applications for spectral CT are in a small animal scanner/specimen spectral CT scanner. Requirement is to mimic material decomposition features of fat/breast tissue/benign and malignant microcalcifications/tumour - but not motion.

"the 'clinical' dose will be slightly higher than the 'phantom' dose. Is there any good reason for that?". This slightly overestimation of the mean glandular dose without the proposed corrections has to be  brought into relation to other uncertainties, e.g. the used kV or better the measured Al-HVL which respects the hardening/roughness of the target and the used filtering or the uncertainty of the accuracy of the used dosemeter. The ISO handbook of  error estimations can help you to perform that calculation.

Without any doubt the depth dose curve is different: see comment of Hanno.

You can find a starndard dataset here:

http://www.ultrasound-images.com/breast.htm#Ultrasound images of lactating breast

As you know these are difficult to obtain. I did find some excellent images on google med.

In general, Screening test is used to detect any speculated patients from healthy-looking subjects. These tests are used to pick up the speculated subjects from any subjects who have a high risk of the target disease (e.g. breast cancer) according to their personal histories. The screening tests usually have a high sensitivity but maybe can give false-positive result

After screening, the diagnosis will confirmed by using the combination of several techniques which have a high specificity and low false-positive as well as false-negative result.

Thus, they are difference in purpose and accuracy.

Hope this can help.

Dear Mercieca,

you can compare following items in your study:

a-The duration of procedures(under local and/or general anesthesia and total time)

b-patient opinion about them:(painstaking,bothering,unacceptable,acceptable,quite acceptable....e.g)

c-Physician's(user's) opinion about easiness of procedures(difficult,moderate and easy)

d-the cost of procedures..

These factors may give better understanding about the discomfort of each modality...

kind regards