Optical mammography

Introduction

The incidence of breast cancer in the UK alone is around 41,000 new cases each year, of whom about one third die of the disease. It is recognised that mortality rate can be reduced by early detection and appropriate treatment. X-ray mammography is the imaging modality of choice in women over 35-40 years of age, but breast ultrasound is used in younger women with denser breasts, due to the reduced effectiveness of x-ray mammography. X-ray mammography also suffers from a significant number of false positives which often lead to unnecessary biopsy. Other conventional diagnostic techniques such as magnetic resonance imaging (MRI) play a significant role in the diagnosis and characterisation of breast disease, but x-ray mammography remains the primary method of detecting and staging breast tumours. Time resolved optical tomography offers several major advantages over existing imaging techniques: a) display of soft tissue contrast based on functional parameters (e.g. blood volume and oxygenation) as well as structural differences; b) use of harmless doses of non-ionising radiation; c) simultaneous reconstruction of the entire three-dimensional volume of the breast; d) no requirement for compression of the breast.

monstirimages


Figure 1
: Source and detector fibres attached
to three rings mounted on a conical phantom.
.

monstirimages


Figure 2
: Fibre holder rings and support
frame for optical tomography of the breast.

In order to evaluate optical tomography on appropriate volunteers and patients a conical fibre holder has been constructed in the form of a series of three connecting rings, shown in figure 1. The rings can be used in isolation, joined in pairs, or all combined, depending on the size and shape of the breast. Three-dimensional images of conical phantoms were successfully reconstructed using a combination of all three rings (click here for more details). For imaging the breast, the rings are supported by an adjustable frame as shown in figure 2. The volunteer leans against the frame with either the left or right breast placed in either of the two holes. The rings within the holes are of different diameters to accommodate a variety of breast sizes. The volunteer sits or stands, with her raised arms and head resting on a cushioned pad mounted on the top of the frame. No compression of the breast is involved. The height of the rings and the angle of the frame are both adjustable.

Two-dimensional imaging

The studies performed so far have involved acquiring scans using single rings of different sizes on both left and right breasts [1]. In each case, rings have contained 16 sources and detectors, and the duration of each scan has been about 10-15 minutes. Initial tests on healthy volunteers enabled the comfort and effectiveness of the fibre holder to be assessed, with positive results. The initial evaluation phase has also indicated that healthy breast images exhibit distributions of absorption and scatter which are quite heterogeneous, and difficult to interpret. We have subsequently performed scans on several subjects with various benign conditions. For example, figure 3 shows an absorption image of a 31-year-old patient with a fibroadenoma at the top right of the image, just 4 mm below the surface. The lesion had been previously identified in an ultrasound scan (figure 4) with dimensions 13 x 10 x 6 mm. However, our image also reveals other features of similar contrast which have not been correlated with known lesions.

monstirimages


Figure 3: A breast image exhibiting
a fibroadenoma at top left.

monstirimages


Figure 4
: An ultrasound image of the same fibroadenoma (top centre) shown in figure 3.

A 48-year-old woman presented at the breast clinic with a breast mass composed of two separate lumps in the right upper outer quadrant. An MRI scan was performed prior to the optical scan, and the lesion was diagnosed as an invasive ductal grade III carcinoma. Optical scans were performed on both her breasts, and figure 5 shows the absorption images obtained. Figure 6 shows the subtraction MRI image, acquired using the contrast agent gadopentetate dimeglumine (Magnavist). The MRI image in figure 6 is shown in a similar orientation to that of the optical images in figure 5. The optical image obtained in figure 5 shows a large absorbing region in the location of the lesion whose position correlates well with the tumour identified in the corresponding MRI image. The high optical contrast is consistent with what would be expected for a highly vascularised malignant lesion. No similar feature is observed in the corresponding absorption image for the left breast, although there is a small region of high contrast near the bottom left. This is possibly an artefact caused, for example, by poor coupling of a detector to the skin, but may also be due to a surface blood vessel.

Our optical imaging group at UCL has been a member of the consortium of research groups funded by the European Commission to explore the clinical utility of optical mammography techniques. For more information about the activities of the OPTIMAMM consortium, please visit its website.

monstirimages



Figure 5
: Absorption images of a) right and b) left breast of patient with malignant lesion in right breast.

monstirimages



Figure 6
: A subtracted MRI image highlighting
a tumour in the right breast.

Monitoring recovery following laser surgery

A further potential application of optical tomography is the assessment and monitoring of breast tissues following surgery and other forms of treatment. At UCL interstitial laser photocoagulation (ILP) is offered as a routine treatment for fibroadenomas. The procedure involves the application of laser energy to the fibroadenoma, which causes heating, coagulative necrosis, and the induction of an inflammatory response which results in resorption of the lump over a period of up to 12 months. A study has been performed to monitor the changes in the optical properties of breast tissue over a twelve-month period following ILP treatment of a fibroadenoma in the breast of a 26-year-old female [2]. Optical imaging scans of the patient were performed one week before and one week after treatment, and immediately after each of four follow-up visits to the breast clinic. A single ring of 16 source-detector pairs was employed, fitted around the breast surrounding a plane in which the lesion was located.

monstirimages


Figure 7: Absorption images reconstructed assuming constant scatter, acquired a) one week before treatment, b) one week after treatment, c) three months after treatment, d) six months after treatment, e) nine months after treatment, and f) twelve months after treatment.

Figure 7 shows the absorption images acquired at a wavelength of 780 nm resulting from each scan, where scatter was held constant. These were reconstructed using measurements of mean photon flight time. The image acquired one week before treatment (figure 7a) exhibits no strongly dominant feature, although the region of higher absorption located just above and slightly to the right of center is consistent with the expected position of the fibroadenoma. One week after surgery, the image (figure 7b) is dominated by a feature almost certainly due to the induction of a significant acute inflammatory response, with its associated increased blood supply (causing high absorption) around the site of the recovering tissue. This is expected to disperse over a period of weeks as the tissue recovers, and as anticipated this region exhibits a large reduction in contrast in the image acquired three months after surgery (figure 7c). However, after six months, while we expected to observe a further decrease, the image displays two dominant features and an overall enhancement in contrast (figure 7d).

To interpret the absorption images presented in figure 5 it is necessary to consider how local changes in scatter are likely to influence their appearance. A decrease in local scattering will cause the measured temporal distribution of photon flight times to become narrower, which also occurs in the presence of a local increase in absorption. Narrowing of the distribution produces a decrease in the mean flight time datatype. As a consequence, these absorption images reconstructed assuming constant scatter will display regions of low scatter as absorption maxima, and regions of high scatter as absorption minima.

monstirimages

Figure 8: Ultrasound images of treated region of breast three months after treatment.

monstirimages

Figure 9: Ultrasound images of treated region of breast six months after treatment.

monstirimages

Figure 10: Ultrasound images of treated region of breast twelve months after treatment.

Following the six-month scan we compared the optical results with corresponding ultrasound scans. Figures 8 and 9 show ultrasound images of the treated region of the breast obtained three and six months after surgery. The ultrasound image acquired at three months shows a single large cyst, which is accompanied by a second smaller cyst after six months. During the third follow-up visit (nine months after surgery) the patient elected to have the larger cyst drained by fine needle aspiration (FNA). The small sample of oily, clear, slightly yellow fluid extracted from the cyst was found to have optical properties consistent with fat. On the following day an optical scan was performed, and the image is shown in figure 7e. The feature corresponding to the cyst observed after 3 months is now gone, and the contrast due to the second cyst (apparent after six months) is reduced. Instead, the most dominant feature in the image occurs at the surface at the position corresponding to the entry of the FNA needle (due to surface bruising). The image acquired twelve months after surgery (figure 7f) has a very similar appearance to that obtained three months earlier, but without the highly absorbing feature at the surface caused by the FNA. There is still evidence of the second cyst just to the left of center. The ultrasound scan (figure 10) confirmed the remains of a very small cyst (the dark area on the left), as well as residual fibroadenoma (on the right) which was no longer palpable. A more detailed description of this case study is available [2].

Three-dimensional imaging

The use of single rings to image the breast has significant limitations: a rigid ring cannot conform to the shape of the breast, and it is difficult to sample the whole breast volume so that lesions (near the chest wall, for example) are not missed. A further disadvantage is that the coupling of light into and out of the breast is usually highly variable, which reduces the reproducibility with which intensity measurements can be achieved. An approach which overcomes most of these problems is that originally adopted by a group at Philips Research Laboratories [3], who used a container filled with a tissue-matching fluid. This has three significant benefits. First, the container can be made of a sufficient size to accommodate a large range of breast sizes and shapes, enabling the entire 3D volume of the breast to be sampled. Second, the coupling of the source and detector optics at the surface is constant and independent of the subject. And third, the external geometry of the reconstructed volume is known exactly, so an accurate model can be generated. With funding from Cancer Research UK we have recently developed a liquid-coupled interface suitable for breast imaging based on a hemispherical cup. The cup, shown in figure 11, has a diameter of 165 mm, which was estimated to be sufficiently large for the majority of woman undergoing x-ray mammography at UCL.

monstirimages

Figure 11: Hemispherical cup developed for 3D breast imaging.

A patient-supporting table has been built as illustrated in figure 12. The cup is attached to a plastic ring secured firmly into an aperture in the table. A channel is cut into the ring, which allows coupling fluid that overflows from the cup to return via plastic tubing to the fluid reservoir. The cup is filled with fluid from below using a peristaltic pump, which then circulates fluid continuously and slowly to sustain a constant level of fluid. A heating element within the fluid reservoir enables the liquid to be maintained at a constant, comfortable temperature. The table is covered with a layer of foam, and a pillow and towels are provided. A preliminary evaluation of the system on phantoms and a healthy volunteer has been published [4], as well as results from a cohort of patients [5]. Several cups of different diameters have been developed to accomodate patients of different sizes.

monstirimages

Figure 12: Breast scanning table and fluid-circulating mechanism.

References

  1. Yates, TD, Hebden, JC, Gibson, AP, Everdell, NL, Arridge, SR, and Douek M (2005): Optical tomography of the breast using a multi-channel time-resolved imager, Physics in Medicine and Biology 50, 2503-2517. Download PDF file.
  2. Hebden, JC, Yates, TD, Gibson, A, Everdell, N, Arridge, SR, Chicken, DW, Douek, M, and Keshtgar MRS (2005): Monitoring recovery after laser surgery of the breast with optical tomography: a case study, Applied Optics 44, 1898-1904. Download PDF file.
  3. Colak, SB, Papaioannou, DG, 't Hooft, GW, Mark, MB, Schomberg, H, Paasschens, JCJ, Melissen, JBM, and Asten, N (1997): Tomographic image reconstruction from optical projections in light-diffusing media, Applied Optics 36, 180-213.
  4. Yates, TD, Hebden, JC, Gibson, AP, Enfield, L, Everdell, NL, Arridge, SR, and Delpy, DT (2005): Time-resolved optical mammography using a liquid coupled interface, Journal of Biomedical Optics 10(5), 054011. Download PDF file.
  5. Enfield, LC, Gibson, AP, Everdell, NL, Delpy, DT, Schweiger, M, Arridge, SR, Richardson, C, Keshtgar, M, Douek, M, and Hebden, JC (2007): Three-dimensional time-resolved optical mammography of the uncompressed breast, Applied Optics 46, 3628-3638. Download PDF file.