X-Ray Sources for Medical and Biopsy Specimen Radiography

What biopsy specimen radiography finds in excised tissue

A specimen goes straight from the patient into a cabinet X-ray system beside the operating table or in the pathology lab. The radiograph answers four questions:

1. Did the target come out? Radiologists mark a lesion before surgery with a wire or a clip, and the radiograph shows whether that marker sits inside the excised piece.
2. Where are the calcifications? Calcium absorbs far more X-ray energy than the surrounding fat and glandular tissue, so deposits appear as bright specks. Eggshell fragments sieved to 53 and 180 µm showed clearly at a 40.7 µm voxel.
3. How close does the target sit to the cut edge? That distance is the margin, and a positive margin sends the patient back for a second operation. Breast conserving surgery accounts for 65.5 percent of breast cancer operations in Australia and New Zealand.
4. Which block should the pathologist cut? A radiograph shows where the abnormality sits, so the pathologist embeds fewer blocks. On mastectomy cases, in-lab imaging cut block use by 10.6 per case.

Microcalcification size and breast micro-CT settings from Croughan et al. 2026; 2D voltages from Langen et al. 2012; lumpectomy micro-CT from McClatchy et al. 2018; 3D X-ray histology from Katsamenis et al. 2019; lung blocks from Sevim et al. 2026; resected lung from Wollatz et al. 2017; and block utilisation from Arudra et al. 2021. Full references at the foot of the page.

Breast, lung, and bone specimens: what changes for imaging

Breast tissue is the low end. Adipose tissue, glandular tissue, and tumour differ by very little, so contrast collapses as the energy climbs. On a 5 cm phantom, 40 kVp separated the fat and glandular substitutes better than 60 kVp or 80 kVp.
Paraffin-embedded blocks sit slightly higher. Wax and adipose tissue absorb almost identically, so that boundary disappears in a block. One system imaged lung blocks at 32 kVp and 220 µA, two others at 50 to 55 kV.
Bone and calcified tissue push you up the range. Cortical bone absorbs far more than soft tissue at every energy, so bone decides whether your source has headroom above the band, not where the floor sits.
Voltage stays low across all of that work. Power does not. Multiply each published voltage by its current and the studies split in two:

We calculated power from the voltages and currents in Sevim et al. 2026, Katsamenis et al. 2019, Wollatz et al. 2017, and McClatchy et al. 2018. Croughan et al. 2026 states 50 W directly.
Specimen size decides your power and tissue decides your voltage, independently. Those two separate choices make specimen imaging a different specification from the industrial inspection on our applications hub, where one dense object drives both.

Specimen radiography in the operating room vs micro-CT in the lab

Run 2D radiography where the answer has to arrive in seconds, micro-CT where it has to be three-dimensional.
The cabinet radiograph answers retrieval in one exposure, which is why the cabinet sits in the operating room. A pathology lab that installed its own cabinet cut turnaround on core biopsies by 1.04 days and on mastectomies by 2.01 days.
Micro-CT answers margins. A single radiograph shows the full depth of the specimen on top of itself, so you see that a calcification sits near an edge but not which face that edge belongs to. Micro-CT reconstructs the whole volume, so the team assesses every face.
Micro-CT costs time. Across eight margin studies, image procurement ran 4 to 30 minutes, with 4 to 10 typical. The Monash setup took 12 minutes for a 5 cm sample at a 40.7 µm voxel.
Accuracy is still developing, and the reason is resolution. The eight pooled margin studies gave a sensitivity of 0.63 and a specificity of 0.78, and the 32-specimen lumpectomy series reported 50 percent sensitivity. Absorption micro-CT in those studies resolved 150 µm to 240 µm, against microcalcifications that start at 53 µm. Those systems miss features smaller than their own resolution, so resolution is the lever, and the source is where resolution starts.
Procurement times and pooled accuracy from Meshkati Yazd et al. 2025; predictive values from McClatchy et al. 2018; scan times from Croughan et al. 2026; and turnaround from Arudra et al. 2021.

What resolution biopsy specimen radiography needs

Specimen imaging splits into two resolution regimes, and each needs a different machine.
For calcifications and margins, you need better than 50 µm. Microcalcifications in a breast phantom measured 53 to 180 µm and appeared clearly at a 40.7 µm voxel. Across the eight pooled margin studies, voxels ran from 4 µm to 200 µm. The Monash setup measured 81 µm at the fat and glandular boundary, the finest among intraoperative setups under 15 minutes.
For tissue architecture, you need about 8 µm. Three-dimensional X-ray histology of a paraffin-embedded lung biopsy ran at an 8.48 µm isotropic voxel over roughly 9.5 hours. Those scans belong in a pathology lab, and the block then goes on to sectioning.
More resolution does not always buy more diagnosis. Specimens from 32 patients imaged at 12 lines per mm and at 2,000 lines per mm showed no significant difference in accuracy. Pick your regime, then specify to it.
Voxel ranges from Meshkati Yazd et al. 2025 and Croughan et al. 2026; 3D X-ray histology from Katsamenis et al. 2019; and film comparison from Langen et al. 2012.

Focal spot size at your operating power

Two numbers decide how sharp a specimen image gets: the detector pixel and the focal spot. Magnification decides which one limits you.
At low magnification, the detector limits you. The Monash setup ran at 1.35 times magnification, giving an effective pixel of 40.7 µm. Against that pixel a spot measured in single microns adds almost no blur, and raising target power from 10 W to 50 W worsened resolution by only 1 µm.
The focal spot sets how far you can magnify. Geometric blur equals the focal spot size multiplied by the magnification minus one, and the effective pixel equals the detector pixel divided by the magnification. Magnifying shrinks the pixel and grows the blur, so the two cross, and past the crossover more magnification costs resolution.
Run those two figures on a 55 µm detector pixel. A 50 µm focal spot crosses over at about 1.7 times magnification and lands near 33 µm. A 4.5 µm spot reaches about 4 times and 14 µm, about 2.4 times finer in the same cabinet. That 2.4 times separates where intraoperative micro-CT resolution sits today, at 150 µm to 240 µm, from where a microfocus source can put it.
So a spot figure has to come with a power figure. The CX30 is the only source in our range with a published curve:

Curpow CX30 data sheet, 19 August 2026.
Where your specimen sits in the 7 W to 50 W power split decides where you sit on this curve. Block and core work at about 7 W holds a spot in single digits. Whole specimen work at 40 W accepts a 60 µm spot, and so runs at low magnification and coarse voxels.
The CX40 carries the smallest spot in our range at 4.5 µm, and the CX20 reaches 5 µm. The source in the 32-specimen lumpectomy series had a 50 µm spot.

Tube voltage and contrast in soft tissue

Soft tissue specimen imaging runs between 20 kV and 55 kV. Contrast comes from small differences in absorption between fat, glandular tissue, and tumour, and those differences shrink as photon energy rises. That is why the Monash group built its spectra to sit mostly between 16 and 30 keV.
Published testing favours a tungsten target. Tungsten produces characteristic X-rays at energies either too low or too high for breast tissue, leaving the bremsstrahlung continuum to do the work. Silver produces a strong peak at 22.16 keV, right in the useful band, so on paper silver should win. It did not. Silver melts at 962 °C, which capped the silver target at 10 W, while tungsten melts at 3,422 °C and ran at 50 W. Unfiltered tungsten at 40 kVp gave the best contrast-to-noise ratio of every combination tested. Flux beat spectrum.
Filtration lost in every configuration tested. Aluminium, silver, and tin filters each reduced image quality against no filter, because each cut flux without shaping the spectrum enough to compensate.
Every Curpow source uses a tungsten (W) target and a beryllium (Be) window, the combination the tungsten result points to. Operating range separates them:

Curpow CX20, CX30, and CX40 data sheets. Specimen band from the studies above.
Two published settings sit below 40 kV: lung tumour blocks at 32 kVp, and 2D breast specimen radiography at 20 to 25 kV. The CX40 is the only source in our range that reaches both. Those 20 to 25 kV settings come from film cabinets running 0.3 mm focal spots, a hundred times larger than microfocus, so treat that end as headroom rather than a requirement today. That floor is room to magnify a specimen radiograph at the voltage where soft tissue has its contrast, which a 0.3 mm tube cannot do.
Spectral band, anode comparison, and filtration from Croughan et al. 2026; 2D voltages from Langen et al. 2012; and lung block settings from Sevim et al. 2026.

Spot position stability for specimen micro-CT

A micro-CT reconstruction assumes the source stayed in one place for every projection. A 5 cm breast specimen scan collected 3,718 projections over 12 minutes, and a lung biopsy scan collected 3,501 over about 9.5 hours. If the focal spot wanders across the target, the projections stop agreeing about where the object is, and the reconstruction blurs by the drift.
Drift matters in proportion to your voxel. At a 150 µm voxel a few microns of movement disappears. At an 8.48 µm voxel, a few microns is most of one. The CX20 data sheet names spot location stability among its key features, and the CX30 data sheet names a highly stable spot position for CT and 3D work. Spot stability drives our micro CT source selection work too.

Choosing between the CX20, CX30, and CX40 for specimen imaging

Curpow CX20, CX30, and CX40 data sheets.
Take the CX40 for blocks, cores, and any magnified design. Paraffin blocks and biopsy cores draw 6.9 W to 7.5 W between 32 kV and 55 kV, inside the CX40’s 8 W ceiling. Its 4.5 µm focal spot magnifies to about 4 times before blur overtakes the detector pixel, and an 8 µm voxel needs that much magnification. The CX40 also reaches 20 kV, the lowest floor in the range, and at 3.6 kg it is the lightest source we build.
Take the CX30 for whole specimens. Whole 5 cm specimens ran at 50 W in both the breast phantom work and the 32-specimen lumpectomy series, and the CX30 40 W model is the flux end of our range. Flux sets scan time rather than deciding whether the scan happens: a 12 minute scan of a 5 cm specimen at 50 W collects the same photons in about 15 minutes at 40 W, inside the 4 to 30 minute window.
Take the CX20 where you want current at the top of the band. Its 250 µA is the most in the range, and its 25 W runs that same 5 cm scan in about 24 minutes.

Lead time and second sourcing

Curpow builds to a six week lead time. For an OEM qualifying a second supplier or replacing a discontinued source, six weeks is the difference between a redesign and a substitution.
Every source ships as tube, high voltage power supply, and controller in one package, with a digital RS-232 interface, automatic conditioning, and diagnostic logging. That is one line item in your bill of materials, not three.

How to specify an X-ray source for a biopsy specimen radiography system

Work through these six in order:

1. Name the lowest voltage you need, not the highest. If any part of your protocol runs below 40 kV, that alone selects the source.
2. Multiply your voltage by your current and specify the watts. Published work splits near 7 W for blocks and cores, 50 W for whole 5 cm specimens. A maximum current on its own tells you little.
3. Work out your magnification ceiling before you argue about focal spot. Find where spot blur crosses your effective pixel. On a 55 µm pixel, a 50 µm spot reaches about 1.7 times, a 4.5 µm spot about 4 times.
4. Ask for the focal spot at your operating power. A data sheet quotes its minimum spot at minimum power. Ask where the spot lands at the watts your exposure needs.
5. Decide whether you are reconstructing. If so, ask about spot position stability, because drift over a few thousand projections costs resolution you paid for.
6. Check the mechanical envelope. Focus to object distance, weight, beam angle, and window orientation decide whether a source fits your cabinet. A side-window configuration exists where a front window will not fit.

What this means for your source selection

Specimen imaging is a low-voltage, high-contrast problem, which inverts the instincts an industrial inspection engineer brings to the job. You are not trying to get through a dense object. You are holding on to the small absorption difference between one soft tissue and another, and that difference survives only at the bottom of the kV scale.
So specify from the specimen outward. A block or a biopsy core asks for about 7 W between 32 kV and 55 kV, magnified hard enough for an 8 µm voxel, and that is the CX40. A whole 5 cm specimen asks for closer to 50 W, and that is the CX30 40 W model. The CX20, CX30, and CX40 all use a tungsten (W) target and a beryllium (Be) window, and all ship in six weeks.
Talk to us about your specimen imaging system, or compare the full range on our products hub.

Sources

1. “Phase-contrast micro-CT for intra-operative breast tumour margin assessment using a microfocus x-ray source and photon-counting detector”. Croughan, Gureyev, Fox, McCormack, Pollock, Jurkschat, Harker, and Kitchen; Monash University, University of Melbourne, Monash Health, and Peninsula University Hospital. arXiv preprint, submitted 10 September 2026. Anthropomorphic breast tissue phantoms and a formalin-fixed paraffin-embedded breast block containing ductal carcinoma in situ, imaged across two anode targets, four filtration options, and four tube voltages.
2. “Micro-computed tomography enables rapid surgical margin assessment during breast conserving surgery”. McClatchy et al.; Breast Cancer Research and Treatment, volume 172, issue 3, December 2018. Prospective imaging of 32 breast conserving surgery specimens, 28 malignant and 4 benign, correlated against final histopathology.
3. “The role of micro-CT in breast cancer management: a systematic review on the clinical applications of micro-CT in breast cancer and a diagnostic accuracy meta-analysis on intraoperative margin assessment”. Meshkati Yazd et al.; Frontiers in Medical Engineering, volume 3, September 2025. Eight of 2,921 retrieved studies pooled for diagnostic indices.
4. “X-ray Micro-Computed Tomography for Nondestructive Three-Dimensional (3D) X-ray Histology”. Katsamenis et al.; University of Southampton; The American Journal of Pathology, volume 189, issue 8, 2019. Formalin-fixed paraffin-embedded human lung biopsy specimens imaged at an 8.48 µm isotropic voxel and returned to conventional sectioning.
5. “3D Histopathology: a Lung Tissue Segmentation Workflow for Microfocus X-ray Computed Tomography Scans”. Wollatz et al.; Journal of Digital Imaging, volume 30, issue 6, December 2017. Surgically resected human lung tissue in wax blocks of about 1.5 cm³, imaged at an 8 µm voxel on a microfocus source with a molybdenum reflection target.
6. “Can micro-CT distinguish between solid lung tumors?”. Sevim et al.; Journal of Pathology and Translational Medicine, volume 60, issue 2, March 2026. Fifteen paraffin blocks, five each of solid adenocarcinoma, non-keratinizing squamous cell carcinoma, and carcinoid tumour, scanned at a 10.1 µm pixel size.
7. “Microradiography of Microcalcifications in Breast Specimen: A New Histological Correlation Procedure and the Effect of Improved Resolution on Diagnostic Validity”. Langen et al.; Radiology Research and Practice, 2012. Specimens from 32 patients, 10 malignant and 22 benign, compared at 12 lines per mm and 2,000 lines per mm.
8. “In-laboratory breast specimen radiography reduces tissue block utilization and improves turnaround time of pathologic examination”. Arudra et al.; BMC Medical Imaging, volume 21, issue 1, March 2021. 144 cases imaged over 48 business days in a pathology laboratory.
9. Curpow CX20 microfocus X-ray source data sheet.
10. Curpow CX30 microfocus X-ray source data sheet, 19 August 2026.
11. Curpow CX40 microfocus X-ray source data sheet.