Cloud DICOM Ingestion: Per-Study vs Per-Gigabyte Pricing
- Study sizes vary from about 15 MB for a chest X-ray to over 2 GB for 3D mammography.
- Per-gigabyte pricing scales with actual cloud infrastructure costs better than a flat per-study fee.
- A flat per-study price forces vendors to overcharge small modalities or lose money on large scans.
- Association retries can make scanners resend an entire study after a connection drop, inflating transfer costs.
- A sustainable pricing model combines a base fee per active endpoint with metered byte volume.
Predicting cloud DICOM ingestion costs depends on one core architectural choice: paying per study or paying per gigabyte. For cloud data transfer, per-gigabyte pricing is almost always the right choice for engineering teams. Study sizes vary from 15 MB for a plain chest radiograph to over 2 GB for 3D mammography. A flat per-study price forces you to overcharge small modalities or lose money on large scans.
Why cloud DICOM ingestion pricing models struggle with study variance
Cloud infrastructure providers do not bill for clinical events. They bill for compute runtime, storage volume, and network egress bytes.
Healthcare business teams often prefer a per-study fee. Hospitals bill patients per examination, so charging per study matches clinical revenue. But treating a medical imaging study as a uniform unit creates financial risk.
In the DICOM standard, a study is a logical folder. Our guide on the DICOM instances, series, and studies hierarchy explains how studies group multiple series. A dental X-ray study contains 2 small files. A trauma chest exam contains 3,000 axial slices.
When your cloud receiver ingests data, network sockets process raw binary streams. Senders transmit files via C-STORE under DICOM Part 4 or via HTTP POST under DICOM Part 18 STOW-RS. The network moves actual bytes, not abstract clinical records. If you price your ingestion per study, your margins will fluctuate wildly with the modality mix of each hospital.
How study size variance across modalities breaks unit economics
Medical imaging modalities produce radically different data volumes. The table below outlines typical payload sizes across common clinical exams:
| Imaging Modality | Typical File Count | Typical Study Size | Network Transfer Profile |
|---|---|---|---|
| Plain X-Ray (CR / DX) | 1–4 instances | 10–30 MB | Fast single-burst transfer |
| Ultrasound (US) | 20–100 frames / loops | 30–120 MB | Mixed single frames and short video clips |
| Magnetic Resonance (MRI) | 200–800 instances | 80–400 MB | Multiple series across scan sequences |
| Computed Tomography (CT) | 800–3,000 instances | 300–1,500 MB | High slice counts, prolonged associations |
| Digital Breast Tomosynthesis (DBT) | 50–200 reconstruction slices | 1,000–2,500 MB | Very dense multi-frame projections |
| Whole Slide Imaging (WSI) | Pyramidal tiles / planes | 1,500–4,000 MB+ | Gigapixel digital pathology data |
Notice the range between modalities. A single 3D mammography exam generates 100 times more data than a digital chest X-ray.
Software teams often try to simplify this variance. They calculate an average across all modalities, such as 150 MB per study. They use this average to set a flat per-study price.
This average fails as soon as customer behavior shifts. If a new hospital site routes full oncology CT scans instead of general outpatient X-rays, your data volume expands tenfold. If you charged a flat €0.50 per study, that fee no longer covers the underlying network transfer.
Should you pay per study or per gigabyte for DICOM ingest?
When evaluating vendor contracts or designing your internal billing, you must choose between these 2 pricing philosophies.
The per-study model
A per-study model charges a fixed fee for every completed examination, regardless of file count or total megabytes.
Advantages:
- Hospital accounting teams understand it immediately.
- Revenue forecasts scale directly with patient exam counts.
- It hides technical networking details from sales conversations.
Trade-offs:
- You absorb all bandwidth risk when facilities adopt high-resolution scanners.
- Senders that submit multi-phase CT or DBT studies erode your gross margin.
- You must either overprice simple X-rays or add complex modality surcharges to prevent losses.
The per-gigabyte model
A per-gigabyte model charges for the actual volume of data transferred through the gateway.
Advantages:
- Your costs scale in direct proportion to your cloud infrastructure bills.
- Plain X-ray clinics pay very little, making your software attractive to small practices.
- Heavy imaging centers pay for the infrastructure capacity they consume.
Trade-offs:
- Monthly invoices vary based on patient volume and imaging protocols.
- You must meter incoming payload bytes per configured endpoint.
We pick per-gigabyte pricing for data pipelines. It reflects the physical cost of moving bits over the internet.
What do you give up? You give up predictable, identical invoices for your finance department. In exchange, you protect your business against margin collapse when handling heavy modalities.
The hidden infrastructure costs of cloud DICOM transfer
Ingesting DICOM into cloud infrastructure involves several distinct operational expenses. Neglecting these items leads to surprise bills.
Cloud network egress
Most cloud providers allow free data ingress. But once bytes enter your cloud environment, moving them between availability zones or out to the internet incurs egress fees. Public cloud providers typically charge between $0.08 and $0.12 per gigabyte for internet egress.
Bytes cross network boundaries when your gateway forwards studies to a downstream webhook. Learn how to handle this in our guide on designing a resilient webhook receiver for DICOMweb STOW-RS. If you ingest 10 TB of imaging monthly, egress alone adds hundreds of dollars in operational cost.
Association retry amplification
Hospital network connections are notoriously unstable. Scanners and PACS archives drop TCP connections frequently. When a transfer fails midway through a 2,000-slice CT scan, standard DICOM senders retry automatically.
Many modality senders do not resume from the failed slice. They restart the entire C-STORE association from slice 1.
If an association drops twice on slice 1,800, the scanner transmits 5 GB for a 1.5 GB study. Under a per-study contract, you collect revenue for 1 study while paying cloud egress for 3 full transfers. Read our guide on handling large DICOM studies for details on managing these exams.
Idle connection overhead
Every connected hospital site requires mutual TLS certificates and network routing resources. Maintaining an open connection endpoint consumes infrastructure capacity even when the hospital sends no studies over the weekend. A sustainable pricing model accounts for active connection endpoints alongside raw byte throughput. For listener architecture tradeoffs, read our guide on building a custom DICOM listener vs a managed gateway.
How to model your monthly DICOM transfer costs
To model your monthly budget, combine a base endpoint fee with metered byte volume:
$$\text{Monthly Cost} = (\text{Active Endpoints} \times \text{Base Rate}) + (\text{Ingested Gigabytes} \times \text{Egress Rate})$$
Consider 2 clinical software companies with very different imaging profiles:
Company A: Urgent care teleradiology platform
- Connected sites: 20 clinics
- Monthly volume: 4,000 studies
- Primary modality: Plain radiography (DX), averaging 25 MB per study
- Total data volume: 100 GB per month
- Bandwidth cost: Low. At €0.15 per GB, data transfer costs roughly €15 per month. The primary cost driver is maintaining the 20 site connections.
Company B: Oncology AI analysis platform
- Connected sites: 4 tertiary cancer hospitals
- Monthly volume: 1,200 studies
- Primary modality: Contrast CT and 3D mammography, averaging 1,200 MB per study
- Total data volume: 1,440 GB per month
- Bandwidth cost: High. At €0.15 per GB, data transfer costs roughly €216 per month. Even with one-third the study count of Company A, Company B consumes over 14 times more network bandwidth.
Suppose both companies used a flat fee of €0.20 per study. Company B would collect €240 while spending almost all of it on cloud bandwidth alone.
Setting up your ingestion pipeline
When designing your cloud medical imaging architecture, align your billing metrics with your infrastructure reality. Build metering into your ingestion layer to track payload bytes per hospital sender.
When you configure your services to receive deliveries, ensure downstream webhooks stream bytes directly to cloud storage. Never buffer full studies in memory.
A per-gigabyte model keeps your platform profitable across all workloads. Your margins hold whether a hospital sends 10 dental films or thousands of dense oncology scans.
Dicomly builds infrastructure for healthcare software teams that receive DICOM studies from hospitals. Why we built Dicomly →