The error reconstruction of real projection for ASD-POCS and FDK in bottom of object model

[PengJinHuang]

the first image is using ASD-POCS and next image is using FDK ,the projections is real data by using CBCT equipment to collect the head model , the third image is one projection , you can see the top of head is in bottom of projection image, I'm comfused why using the ASD-POCS to iterative the reconstruction result ,the bottom of result is so terrible, although the FDK also lead to a bad result in bottom of reconstruction image because of cone beam

[AnderBiguri]

@PengJinHuang Good question!

Iterative algorithms like ASD-POCS do really bad when the data is inconsistent. It is plausible that your projection data is overattenuated in the edges, and that, together with the cone-beam artifacts that are generally expected (as FDK), its making that part of the image really bad. there are data tricks to minimize this, which I can help with (send me an email), but I am quite busy until the end of the month.

[PengJinHuang]

@PengJinHuang Good question!

Iterative algorithms like ASD-POCS do really bad when the data is inconsistent. It is plausible that your projection data is overattenuated in the edges, and that, together with the cone-beam artifacts that are generally expected (as FDK), its making that part of the image really bad. there are data tricks to minimize this, which I can help with (send me an email), but I am quite busy until the end of the month.

First of all, thank you for your help despite your busy schedule , im not in a hurry to receive the result and i have sended the projections and corresponding projections collection informations, and i find that the TIGRE presume the line which connect the center of img volume and center of detector panel will be perpendicular to the panel when you use default geo, because i use different projections which the equipment may have different degrees of mechanical correction ,the asd-pocs for some projections will lead to a apparent geometric artifact, ,maybe it can be resolved by using geometric calibration but i have no time to do it ,but dont warry, the projections i send to you have been calibrated to strive to confirm the ideal conditions as much as possible.the reconstruction result is show on top.and another question about you say is overattenuated projection in the edges ,projections is intensity of xray after xray through the object, why is overattenuated in bottom ? it's because of hardening ?I didn't understand what over attenuation means here

[amirahsp]

I think this over attenuation caused by the gantry table or stand where you put the head on it! So the level of signal tends to zero in the bottom of projections.

[AnderBiguri]

@amirahsp that could be an issue, as for FDK its not a problem, but for iteratives it is. Its better sometimes to crop that area

[AnderBiguri]

I'll close the issue, but let me know if you need more help and I'll reopen