What this page shows, in plain English
Tap any heading to expand it.
First, what the septum is
The wall running down the middle of your nose that separates the left passage from the right. It is part bone at the back and part cartilage at the front.
The big colour map
Picture the septum as a flat sheet standing on edge inside your nose. This map is that sheet, unfolded and viewed side-on. Left-to-right is front-of-nose to back-of-nose. Up-and-down is floor to roof. The colour at each spot says which way the wall leans there, and by how many millimetres — one colour for leaning left, the other for leaning right, pale for straight.
Why there is no single headline number
People say “deviated septum” as though it were one thing with one value. This map shows the wall bending different ways in different places — leaning one way up high and the other way further down and back. Collapsing that into a single number would throw away the actual shape, so the page deliberately never does.
The line charts
The same information drawn as simple lines: how far off-centre the wall gets as you move further back, and as you move higher up. The third chart is the bend angle — how sharply the wall curves rather than how far it has moved.
The trace check picture
The path the computer followed, drawn on top of real cross-sections, so you can confirm it actually tracked the septum. This is not decoration: earlier attempts followed the turbinate, the sinus wall, and the wall between the sphenoid sinuses instead, inventing dramatic false readings each time. Looking at this picture is how those were caught and fixed.
The grey and red patches
Spots where the computer could not confidently find the wall — usually because it is very thin there, or fused with the tissue beside it. They are left visibly blank rather than quietly filled in with a guess.
Where this is least reliable. CT shows bone clearly but cartilage poorly, and the front of the septum is mostly cartilage — so the front of the map is the softest part of it. What is traced there is the middle of the whole soft-tissue wall, lining included, and a swollen lining alone could shift it a couple of millimetres on a different day.
NOT A MEDICAL DEVICE · NOT A DIAGNOSIS · one person's CT, explored for personal understanding
Septal geometry explorer
GE Revolution Maxima · PNS-ORBIT-TMJ helical, 120 kVp · Series 3 (bone kernel,
0.588 mm px, 0.625 mm slices) · analysed 9,133 septum positions over a
86 mm (depth) × 59 mm (height) sheet
"The deviation" of a nasal septum is not one number. The septum is a surface, and how far
it sits from the midline changes with where along it you look and at what height.
This page shows the whole surface at once, so no single figure has to stand in for it.
On this scan the traced partition ranges from -7.2 mm (toward the patient's right,
at depth 30 mm / height 58 mm) to
+8.5 mm (toward the left, at depth 76 mm /
height 48 mm) — both directions, on the same septum.
No claim is made here about whether any of this is significant; that is a clinical question
this page deliberately does not answer.
The displacement map
blue = sits toward patient RIGHT
orange = sits toward patient LEFT
grey = gap: partition not resolved, left blank (never interpolated)
Check the trace yourself
Every measurement above depends on the cyan line actually following the septum.
Slide through the coronal reformats (built directly from the axial volume — the scanner's own
reformats have uneven slice spacing and were not used) and judge for yourself.
Gold line = fitted midline; red squares mark gap heights.
Profiles and bend angle
What the midline even is
No midline exists in the data by itself; one has to be defined. Here it is the plane that best
maps the bony skull onto its own mirror image (teeth/mandible excluded), refined over lateral
position and two tilt angles. The fit landed 3.3 mm from the array centre with a
2.2° head yaw — assuming the scanner grid was the midline would have
biased every number on this page. The plane passes within 0.4 mm of the crista galli
(the stable bony crest above the septum), an independent check.
| midline tilted by | mean shift | 95th pct | extremes become |
|---|
| yaw+1° | 0.98 mm | 1.59 mm | -9.0 / +9.0 mm |
| yaw-1° | 0.98 mm | 1.59 mm | -6.4 / +10.0 mm |
| roll+1° | 0.31 mm | 0.66 mm | -7.4 / +9.6 mm |
| roll-1° | 0.31 mm | 0.66 mm | -7.9 / +9.4 mm |
The whole analysis was re-run four times with the midline plane deliberately tilted
1° each way. That is the honest uncertainty from the midline choice alone.
Caveats that belong on the same page as the pictures
Cartilage is nearly invisible to CT. The bony septum (vomer, perpendicular
plate) and the cartilaginous septum behave differently, and CT shows cartilage poorly. The anterior
~2–3 cm of the septum is largely cartilage, so the front of this map — often the part that
matters most for airflow — is exactly where the trace is least reliable. A soft-kernel
cross-trace of the anterior 25 mm agreed to 0.2 mm on average
(95th pct 0.4 mm) at 411 test points, which bounds but does
not remove that doubt.
Gaps and holes are real and left visible. 73 septum positions
(0.8%) had a partition too thick to call a septum (fused with a turbinate, mucosal
bridging) and are shown as grey/red squares; a further 371 positions (3.9%
of the sheet) had no air to measure against on one side (a completely blocked slit) and are simply
blank. Nothing anywhere is filled in by interpolation.
The trace stops at the choana. Columns at the very back, where the two nasal
cavities have already merged below the partition, were cut off — what remains up there is the
sphenoid inter-sinus septum, a different structure that would otherwise masquerade as a large
posterior "deviation". Near the roof, the tracer can also grab the wall of an ethmoid air cell
adjacent to the perpendicular plate instead of the plate itself, so the top ~10 mm of the map is
the noisiest band.
Mucosa is included in "the partition". The trace finds the whole soft-tissue
wall between the two air spaces; swollen mucosa (which changes hour to hour with the nasal cycle)
moves the traced centre without the septum itself moving.
Reproduce with make_septum_geometry.py · raw trace in
septum.csv · known weaknesses in SOFT_SPOTS.md
